GB2489035A

GB2489035A - OFDM channel estimate interpolator which calculates virtual replacement pilot at discontinuities in scattered pilot pattern

Info

Publication number: GB2489035A
Application number: GB1104534.1A
Authority: GB
Inventors: Matthew Paul Athol Taylor
Original assignee: Sony Corp
Current assignee: Sony Corp
Priority date: 2011-03-17
Filing date: 2011-03-17
Publication date: 2012-09-19
Also published as: GB201104534D0

Abstract

Orthogonal frequency division multiplex (OFDM) systems typically use scattered pilot training symbols for channel response estimation, from which channel responses of data sub-carriers are interpolated. These are typically arranged in regular repeating patterns, however there can be discontinuities in these patterns (eg. at a frame boundary, see Figs. 3,5). The interpolator comprises an FIR filter 214.1 which operates on the expected 212 and received 208 received symbols. The filter requires the regular repeating pattern for correct operation. Where a break in the pattern causes an irregularity the controller 320.1 causes the re-sample generator 324.1 to create virtual pilots to maintain the pattern. The new pilot sample location 466 is kept and the older samples 464 are discarded and replaced with calculated pilot samples 470 in the correct pattern position (see Fig. 10d). The invention may be applied in DVB-T2 systems.

Description

RECEWER AND METHOD

Field of the invention

The present invention relates to receivers and methods for receiving Orthogonal Frequency Division Multiplexed (OFDM) symbols, at least some of the OFDM symbols including a plurality of data bearing sub-carriers and a plurality of pilot bearing sub-carriers.

Background of the invention

There are many examples of communications systems in which data is communicated using Orthogonal Frequency Division Multiplexing (OFDM). Systems which have been arranged to operate in accordance with Digital Video Broadcasting (DVB) standards for example, use OFDM. OFDM can be generally described as providing K narrow band sub-carriers (where K is an integer) which are modulated in parallel, each sub-carrier communicating a modulated data symbol such as Quadrature Amplitude Modulated (QAM) symbol or Quadrature Phase-shift Keying (QPSK) symbol. The modulation of the sub-carriers is formed in the frequency domain and transformed into the time domain for transmission. Since the data symbols are communicated in parallel on the sub-carriers, the same modulated symbols may be communicated on each sub-carrier for an extended period, which can be longer than a coherence time of the radio channel. The sub-carriers are modulated in parallel contemporaneously, so that in combination the modulated carriers form an OFDM symbol. The OFDM symbol therefore comprises a plurality of sub-carriers each of which has been modulated contemporaneously with different modulation symbols.

To facilitate detection and recovery of the data at the receiver, the OFDM symbol can include pilot sub-carriers, which communicate symbols which are known to the receiver. The pilot sub-carriers provide a phase and timing reference, which can be used to estimate an impulse response of the channel through which the OFDM symbol has passed, to facilitate detection and recovery of the data symbols at the receiver. In some examples, the OFDM symbols include both Continuous Pilot (CP) carriers which remain at the same relative frequency position in the OFDM symbol and Scattered Pilots (SP). The SPs change their relative position in the OFDM symbol between successive symbols, providing a facility for estimating the impulse response of the channel more accurately with reduced redundancy.

In co-pending UK patent application number GB0909579.5 there is disclosed a receiver for receiving a sequence of OFDM symbols transmitted via a channel, and in particular a receiver for receiving OFDM symbols, which have been transmitted in accordance with DVB-T2. Each OFDM symbol comprises a plurality of data bearing sub-carriers on which data is transmitted and a plurality of pilot bearing sub-carriers on which pilot data is transmitted. The pilot sub-carriers are distributed throughout the OFDM symbols of the sequence in accordance with a predetermined pilot sub-carrier pattern. The receiver includes a channel estimator, and the channel estimator includes a pilot data extractor for extracting pilot data from the pilot sub-carriers of each OFDM symbol; a pilot data extrapolator for generating extrapolated pilot data based on pilot data extracted from the pilot data sub-carriers; and a pilot adjusting unit operable to process the pilot data by interpolating between the extrapolated pilot data in time and frequency to produce an estimate of the channel. The receiver also comprises a discontinuity detector for detecting a discontinuity in the pilot data processed by the channel estimator, and a controller, which upon detection of a pilot data discontinuity by the discontinuity detector, is operable to provide a control signal to the channel estimator which causes at least one of the pilot data extractor, the pilot data extrapolator and the pilot adjusting unit to compensate for the pilot data discontinuity in the pilot data. As such the receiver can be arranged to receive data from OFDM symbols, such as DVB-T2, which include a number of features which may give rise to discontinuities in pilot data extracted at the receiver. In order to accommodate a resulting discontinuity, in the pilot data symbols caused by a discontinuity the receiver is arranged to detect pilot data discontiriuities at the receiver and a controller is provided for ensuring that upon detection of a discontinuity in the pilot data, at least one part of the channel estimator is adapted so as to accommodate for the discontinuity.

However it is envisaged that there is a requirement for further improvements in receivers which must cope with diseontinuities in pilot sub-carriers.

Summary of Invention

According to an aspect of the present invention there is provided a receiver for receiving data from a sequence of OFDM symbols transmitted via a channel, each of the OFDM symbols comprising a plurality of data bearing sub-carriers on which the data is transmitted and a plurality of pilot symbol bearing sub-carriers on which pilot symbols are transmitted. The pilot sub-carriers are arranged within the OFDM symbols in accordance with a predetermined pilot sub-carrier pattern. The receiver includes a channel equaliser comprising a pilot symbol extractor for extracting the received pilot symbols from the pilot sub-carriers of each of the received OFDM symbols, a channel estimator which is arranged in operation to form, for each pilot sub-carrier, an estimate of a sample of the channel through which the received OFDM symbols have passed by comparing the received pilot symbols with a version of the pilot symbols transmitted with the OFDM symbols, and to interpolate the samples in time andlor frequency to produce an estimate of the channel, and an equaliser. The equaliser is arranged in operation to reduce the effects of the channel on the received OFDM symbols using the estimate of the channel generated by the channel estimator, so that data can be recovered from the received OFDM symbols. The channel estimator includes an interpolator which is adapted to detect a change in a relative sampling of the channel derived from a change in an interval between the pilot symbol sub-carriers which have been received and a newly received pilot sub-carrier symbol, to determine an adapted relative position of the samples of the channel provided by the pilot symbol sub-carriers which have been received to produce a relative position of the samples of the channel which corresponds to a sampling position for the newly received pilot symbol sub-carrier, to re-calculate the samples of the channel estimate in accordance with the adapted relative position of the samples, and to use a sample of the channel estimate derived from the newly received pilot symbol sub-carrier in combination with the re-calculated samples of the channel estimate to generate the interpolated samples of the channel estimate.

A receiver for detecting and recovering data from OFDM symbols uses a channel equaliser to generate an estimate of the channel through which the OFDM symbols have passed using pilot sub-carrying symbols which are distributed throughout the OFDM symbols, and which change from one OFDM symbol to the next. For example the pilot symbol carrying sub-carriers may include scattered pilots as explained above. As explained above, a location of at least some of the pilot symbol bearing sub-carriers, for example the scattered pilots, within each OFDM symbol changes in accordance with a predetermined pattern. The pattern is arranged such that a different location of the pilot symbols causes a distribution of samples of the channel to be produced in the frequency domain.. The separation of the samples of the channel produced by the pilot symbol sub-carriers in the frequency domain Dx therefore provides a sampling bandwidth in frequency. Furthermore, in accordance with the predetermined pattern, the location of the pilot bearing sub-carriers will repeat, so that a pilot symbol will occur at the same position within an OFDM symbol after a predetermined number of OFDM symbols Dy. Therefore by interpolating in time as well, as frequency a sample of an estimate of the channel can be produced for each data bearing sub-carrier of the OFDM symbols. As such, the effect of the channel therefore be reduced or cancelled from the received OFDM symbols with an effect that the data can be recovered from the OFDM symbols with an improved likelihood of being correct. However, a technical problem may be presented when there is a change in a sampling pattern provided by the predetermined pattern of pilot symbols. Although the predetermined pattern remains consistent and repeating to generate the sampling effect required, the pattern may change. In one example the pattern changes because a structure of frames within which the OFDM symbols are transmitted may not align with a repeating pattern of the pilot symbol sub-carriers. As such, for example, where one frame changes to the next, the pilot pattern of the next frame may not be consistent with the pattern from the previous frame, as the pattern re-starts. As a result there may be a change in the relative sampling rate or interval between samples.

Embodiments of the present technique can provide an equaliser with.in a receiver for OFDM symbols, which includes an interpolator for generating samples of a channel estimate for use in equalising the received OFDM symbols which can accommodate a change in the sampling pattern produced by a change in a regularity of a pattern of pilot symbol sub-carriers, with an effect that a discontinuity in the

S

sampling of the channel which is produced may reduce an error in the samples of the channel estimate generated by the interpolator.

In some examples the change in a relative location of the sampling rate produced by the pilot symbol sub-carriers may be caused by a change in the pilot S symbol sub-carrier pattern from one frame to the next.

In one example, the OFDM symbols are transmitted in accordance with a DVB standard such as DVB-T2.

Various further aspects and features of the invention are defined in the appended claims, which include a method of receiving.

Brief Deicri! tion of the Drawin Embodiments of the present invention will now be described by way of example only with reference to the accompanying drawings where like parts are provided with corresponding reference numerals and in which: Figure 1 provides a schematic diagram showing a typical DVB-T2 transmitter chain; Figure 2 provides a provides a schematic diagram showing a typical DVB-T2 receiver chain; Figure 3 provides a schematic diagram showing a sequence of OFDM symbols, which are generated for example according to the DVB-T2 standard, and which illustrate one example of a predetermined pattern of pilot symbols carrying sub-carriers; Figure 4a is a schematic block diagram of a channel estimation and correction unit with an FR-interpolator which forms part of the receiver shown in Figure 2; Figure 4b is a schematic block diagram of a channel estimation and correction unit with an extrapolator which forms part of the receiver shown in Figure 2; Figure 5 provides graphical illustration of samples of a channel estimate which are formed from pilot symbols received on sub-carriers of OFDM symbo]s corresponding to the position of these sampled, according to an example predetermined pilot pattern; Figure 6a is a schematic block diagram of an interpolator formed using an Finite Impulse Response (FIR) filter; and Figure 6b is a schematic block diagram of an interpolator which is adapted in accordance with the present technique; Figures 7a to 7c provide graphical illustrations of samples of a channel estimate which are formed from pilot symbols received on sub-carriers of OFDM symbols which are interpolated by the interpolator shown in Figure 6a; Figures 8a to 8e provide graphical illustrations of samples of a channel estimate which are formed from pilot symbols received on sub-carriers of OFDM symbols which are interpolated by the interpolator shown in Figure 6b; Figures 9a to 9e provide graphical illustrations of samples of a channel estimate showing an example operation of the interpolator shown in Figure 6b; Figures lOa to 1 Oi provide graphical illustrations of samples of a channel estimate showing another example operation of the interpolator shown in Figure 6b; Figure 11 provides a graphical illustration of two sample plots of a channel estimate for an example set of samples before the operation of the interpolator of Figure 6b and after, for the case where a sample of the channel from the next OFDM symbol is inegularly spaced but close to a final sample of a present OFDM symbol; Figure 12 provides a graphical illustration of two sample plots of a channel estimate for an example set of samples before the operation of the interpolator of Figure 6b and after, for the case where a sample of the channel from the next OFDM symbol is irregularly spaced but close to a final sample of a present OFDM symbol; Figures 13a to l3m provide graphical illustrations of samples of a channel estimate showing another example operation of the interpolator shown in Figure 6b; Figure 14 is a schematic block diagram of an interpolator, which includes a extrapolator according to a previously proposed arrangement; Figure 15 provides graphical illustration of samples of the channel estimate corresponding to the example shown in Figure 5 illustrating the operation of the interpolator shown in Figure 14 according to one example; Figure 16 provides graphical illustration of samples of the channel estimate corresponding to the example shown in Figure 5 also illustrating the operation of the interpolator shown in Figure 14; Figure 1 7a provides graphical illustration of samples of the channel estimate corresponding to the example shown in Figure 10 illustrating how the operation of the interpolator shown in Figure 14 according to the present technique when a gap between samples of the channel estimate is greater than. a displacement at a sampling rate provided by the pilot pattern causes a problem; Figure 17b provides an example of the operation of the interpolator when the gap between the samples of the channel estimate is greater than the sampling period of the pilot pattern; and Figure 17e is a second example of the operation of the interpolator when the gap between the samples of the channel estimate is greater than the sampling period of the pilot pattern; and Figure 18 is a flow diagram illustrating a process performed by the interpolator when operating in accordance with the present technique.

Description of Exam&c Embodiments

OFDM Transmitter and Receiver Figure 1 provides an example block diagram of an OFDM transmitter which may be used for example to transmit video images and audio signals in accordance with the DYB-T2 standard. in Figure 1 a program source 1 generates data to be transmitted by the OFDM transmitter. A video coder 2, and audio coder 4 and a data coder 6 generate video, audio and other data to be transmitted which are fed to a program multiplexer 10. The output of the program multiplexer 10 fonns a multiplexed stream with other information required to communicate the video, audio and other data. The multiplexer 10 provides a stream on a connecting channel 12.

There may be many such multiplexed streams which are fed into different branches A, B etc. For simplicity, only branch A will be described.

As shown in Figure 1, an OFDM traiismittei icceives the strewn at a multiplexer adaptation and energy dispersal block 22. The multiplexer adaptation and energy dispersal block 22 randoinises the data and feeds the appropriate data to a forward error correction encoder 24 which performs error correction encoding of the stream. A bit hitei-leavcr 26 is provided to interleave the encoded data bits which for the example of DVB-T2 is the LDCP/BCH encoder output. The output from the bit interleaver 26 is fed to a bit into constellation mapper 28, which maps groups of bits onto a constellation point of a modulation scheme, which is to be used for conveying the encoded data bits. The outputs from the bit into constellation mapper 28 are constellation point labels that represent real and imaginary components. The constellation point labels represent data OFDM symbols formed from two or more bits depending on the modulation scheme used. These can be referred to as data cells.

These data cells arc passed through a time-interleaver 30 whose effect is to interleave data cells resulting from multiple LDPC code words.

The data cells arc received by a frame builder 32, with data cells produced by branch B and C in Figure 1, via other channels 31. The frame builder 32 then forms many data cells into sequences to be conveyed on OFDM symbols, where an OFDM symbol comprises a number of data cells, each data cell being mapped onto one of the sub-carriers. The number of sub-carriers will depend on the mode of operation of the system, which may include one of 1k, 2k, 4k, 8k, 16k or 32k, each of which provides a different number of sub-carriers according, for example to the following table: Mode Sub-carriers 1K 853 2K 1705 4K 3409 8K 6913 16K 13921 32K 27841 Table I: -Maximum Number of Sub-carriers per mode.

The sequence of data cells to be carried in each OFDM symbol is then passed to the OFDM symbol interleaver 33. The OFDM symbol is then generated by an OFDM symbol builder block 37 which introduces pilot and synchronising signals fed from a pilot and embedded signal former 36. An OFDM modulator 38 then forms the OFDM symbol in the time domain which is fed to a guard insertion processor 40 for generating a guard interval between OFDM symbols, and then to a digital to analogue converter 42 and fmally to an RF amplifier within an RF front end 44 for eventual broadcast by a connection to a network cable 46.

As shown in Figure 2 an OFDM signal is received by an antenna 100 and detected by a tuner 102 before being converted into digital form by an analogue-to-digital converter 104. A guard interval removal processor 106 removes a guard interval from a received OFDM symbol, before modulation symbols representing the data are recovered from each received OFDM symbol using a Fast Fourier Transform (FFT) processor 108 in combination with a channel estimator and corrector 110 and an embedded-signalling extraction (decoding) unit 111. The modulation symbols are fed to a frequency de-interleaver 112, which performs a reverse mapping between the modulation symbols and the OFDM symbol sub-carriers to form a stream of modulation symbols from each of the OFDM symbols. A frame de-mapper 114 then separates the modulation symbols transmitted in different frames of the time division multiplexed structure of the OFDM transmission interface into logical channels, which are then time dc-interleaved by a time de-interleaver 115 and then a further deinterleaver called a. cell de-intealeaver 116. A cyclic delay removal unit 117 then to removes a cyclic shift if introduced into the data at the transmitter. The demodulated data is then recovered from a dc-mapper 118 from the modulation symbols and to produce for each channel a bit stream. A bit de-interleaver 120 then reverses any bit interleaving in the signal. Finally an error correction decoder 121 is arranged to correct errors and recovbrs an estimate of the source data.

Also shown in Figure 2 is a control unit 119 which receives information from the output of the error correction decoder 121. As will be appreciated for those familiar with the DVB-T2 Standard, certain information is communicated via signalling symbols referred to as P1 and P2 symbols which can help the receiver to synchronise to a current phase and timing of a time divided frame or a time slicing frame with respect to which different data streams are multiplexed. Furthermore, as will be explained shortly,the control unit 119 detects a current phase of a pilot pattern which repeats in accordance with a predetermined arrangement of pilot symbols. By determining a current phase of the pilot pattern it is possible to identify which of the sub-carriers of the received OFDM symbol are pilot sub-carriers and correspondingly which type of pilot symbols the pilot sub-carriers are carrying.

DV]3-T2 OFDM Symbols As will be appreciated by those familiar with DVB-T2 the pilot pattern is not fixed but changes from symbol to symbol in accordance with a predetermined pattern.

This is because the pilot sub-carriers in successive OFDM symbols include both fixed pilot symbols (CP) and scattered pilot symbols (SP). The scattered pilot symbols change theft relative location within an OFDM symbol from one symbol to another.

An example of a sequence of OFDM symbols in accordance with the DVB-T2 Standard is illustrated in Figure 3.

Figure 3 provides an illustrative representation of a frame structure for DVB-T2. In Figure 3 each of the squares represent "a cell" which is the term used for a data symbol carried by a sub-carrier of an OFDM symbol. As such in Figure 3 each of the rows represents a single OFDM symbol, so that from left to right each of the cells forms part of a single OFDM symbol. Each of the columns therefore represents cells transmitted on the sub-carriers of successive OFDM symbols in time.

II

There are four types of pilots in DVB-T2: Continual, scattered, edge and P2 pilots: o The continual pilots are at fixed carrier positions in the OFDM spectrum and are transmitted every symbol. These are typically used for common phase error (CPE) and time/frequency offset correction * The scattered pilots change in accordance with a predetermined pattern and are used for channel estimation and equalization.

o The edge pilots are at the lowest frequency and highest frequency carrier positions in the OFDM spectrum. They are continuous pilots.

They also used for channel estimation and equalization.

* The P2 pilots are continuous pilots which are transmitted in a special symbol at the start of a frame of symbols, which is used for synchronization and initial acquisition by a receiver.

In Figure 3 a sequence of OFDM symbols is shown with a pilot pattern for a first frame n-i 120 followed by a subsequent frame n 122. Five symbols from the first frame 120 are shown with a pattern of pilot symbol sub-carriers, which changes for a subsequent frame n 122 at a frame boundary 124. Thus, as a result of the change of frame at a boundary 124, the relative sampling position of the channel provided by the pilot sub-carriers changes with respect to the previous frame n-i 120, which can cause a discontinuity in the rate of sampling the channel estimate provided by the pilot symbols.

In the following description of example embodiments the use of expressions relating to a relative position of pilot symbol carrying sub-carriers and a relative position of a sample of the channel estimate which is produced at this relative position may be used interchangeably. This is because embodiments of the present technique are arranged to address a technical problem associated with a change in sampling position or sampling rate caused by a change in a relative rate of occurrence of the pilot symbols within the OFDM symbols.

As shown in Figure 3, the scattered pilot sub-carriers change their relative position in the OFDM symbols from one symbol to the next, so that in accordance with a repeating pattern, within a predetermined number of OFDM symbols of a frame, a regularly spaced sample of the channel is provided. The spacing of the channel samples which is provided by the pattern of scattered pilots is less than a spacing which is provided for any one OFDM symbol. Thus by generating an estimate of the channel using a sequence of OFDM symbols over the repeating cycle a bandwidth of the channel estimate can be much greater than that which can be S achieved from one OFDM symbol. Accordingly using time and frequency interpolation of the samples of the channel acquired by the scattered pilots as well as the continuous and P2 pilots an eslimate of the channel can have a much greater bandwidth. The repeating pattern of the scattered pilots can be referred to as a cycle which comprises a number of phases, each phase specifying the location of the scattered pilots according to the predetermined pattern. A separation in time between pilot symbols before the cycle again produces a scattered pilot at that same location is referred to as Dy, whereas a separation in frequency between one scattered pilot in one OFDM symbol and the scattered pilot in the next OFDM symbol is Dx. In Figure 3, Dy 4 and Dx 3.

Pilot Sub-carrier Pattern As shown in Figure 3, 14 successive OFDM symbols are shown which each include 36 sub-carriers. Thus in Figure 3 the Y axis represents time and cach row of the matrix of squares represents a different OFDM symbol whereas the X-axis represents frequency with each square representing a different sub-carrier. As shown in Figure 3 one of the sub-carriers provide an edge pilot whose the position remains the same in each of the OFDM symbols. A preamble of five symbols in a sequence of Nsym OFDM symbols is shown for the frame n-i 120 followed by a P2 symbol at the boundary 124 which is transmitted at the start of frame n 122. Following the P2 symbol eight successive OFDM symbols are shown. As shown in Figure 3 the position of the scattered pilot SP changes from one symbol to the next so that by accumulating the pilot sub-carriers for successive OFDM symbols a greater sampling of the channel in the time and frequency domain can be achieved than would be possible if the pilot symbols were all transmitted on the same sub-carriers. However as a consequence of the frame structure and the repeating pilot pattern, where for example the boundary 124 occurs between one frame n-I 120 and a next frame n 122, there is caused a discontinuity in the sampling of the channel through which the OFDM symbols have passed. This is because at a boundary 124 between frame n-i and frame n a pilot sub-carrier symbol is located in a position in the time domain which is between locations at which a time interpolator would expect to receive a subsequent sample. More explanation will be provided of this effect shortly.

Conventional Epualiser Structure As will be appreciated in order to generate a channel estimate from the pilot symbol pattern shown in Figure 3 interpolation is performed in both the time and frequency domains. Thus an example of a channel estimator which is used for providing an estimate of the channel for equalisation is shown in Figures 4a and 4b.

Figures 4a and 4b provide example block diagrams illustrating components which provide one example of a channel estimation and correction processor ll 0 which is shown in Figure 2. In Figure 4a for example a frequency domain version of the received OFDM symbols are fed on a connecting channel 200 to a pilot extractor 202. The pilot extractor 202 operates to separate the data bearing sub-carriers 1J14 froni the pilot symbol sub-carriers Zik and feeds respectively the data bearing sub-carriers on a channel 204 to a divider 206 via a compensating delay circuit 205 and a channel 208 to a further divider 210. The pilot symbols Z1 are received on a first input of the divider 210 which receives on a second input 212 a version of the pilot symbols which is known to the receiver in a form in which they were transmitted at the transmitter. Thus by dividing the pilot symbols received at the receiver Z/k with a reproduced version of the pilot symbols when transmitted C, a sample of the channel H, is provided for a position in time and frequency 4k at which the pilot symbol was transmitted by the pilot symbol carrying sub-carrier. The sample of the channel estimate provided by the pilot symbols is then fed to a time interpolator 214 which interpolates the samples of the channel estimate provided by the pilot symbols in time followed by a frequency interpolator 216 which interpolates between the samples of the channel estimate in frequency. Thus at an output of the frequency interpolator there is provided a sample of the channel for each of the data symbols JJ for the received OFDM symbol. Thus collectively the divider 10, the time interpolator 214 and the frequency interpolator 216 form a channel estimator 218. The compensating delay 205 introduces a delay into the data bearing modulation symbols to match a delay caused by the channel estimator 218, so that the division made by the divider 206 of the modulation symbols by the samples of the channel estimate correspond correctly time with each other. The channel estimation and correction processor 110 generates at an output of the divider 206 an equalised signal b/k by dividing the data bearing sub-carriers by the estimate of the channel at respective corresponding sub-carrier locations to form the cqualised signal Ak Figure 4b differs from Figure 4a in that there is no compensating delay circuit 205. This example corresponds to the ease where the time interpolation circuit 214 does not introduce a delay in the estimate of the channel, such as for example in the case where an extrapolator is used for the time interpolator 214 as for example explained below.

the arrangement of the channel estimation and correction processor 110 as shown in Figures 4a and 4b correspond to conventional arrangements for equalising the received UFDM symbols using an estimate of the channel generated from the received pilot symbols. As will be appreciated in order to generate an estimate of the channel for each of the data bearing sub-carrier positions within each of the OFDM symbols then the time interpolator 214 must interpolate the samples of the channel estimate derived from the pilot sub-carriers in time and the frequency interpolator 216 must interpolate between the samples in frequency. Thus the number of pilot symbols transmitted provides a certain sampling bandwidth with respect to which interpolation can be performed, the bandwidth of the sampling being arranged to allow a channel to be estimated accurately in accordance with a respective rate of change of the channel with frequency determined by Nyquist theorem.

Time Interpolation Filter Embodiments of the present invention provide an improved way of performing time interpolation between the samples of the channel provided by the pilot symbol sub-carriers. More particularly but not exclusively, the time interpolation performed by the time interpolator 214 is adapted and arranged to accommodate a change in a relative phase of the pilot symbols as a result of a pilot sub-carrier pattern which causes a discontinuity between the samples of the channel provided by the pilot sub-carriers for example as a result of a change from one frame to the next.

Thus referring to Figure 3 and as explained above as a result of the end of one frame and the start of the next, the insertion of the P2 symbols which provide pilot sub-carrier symbols in time which cause an irregular spacing between the pilot sub-carriers a discontinuity in the sampling of the channel is explained with reference to Figure 5. As shown in Figure 5 vertical lines 250 represent a relative location of sub- carrier symbols of an OFDM symbol with respect to time. A first set of six pilot sub-carriers are shown which generate samples of a channel impulse response represented by a solid line 252. Thus the pilot symbols 251 sample the channel through which the OFDM symbol has passed as a relative spacing of four OFDM symbols. However as represented by a dashed line 124, which illustrates a change from one frame n-I to the next frame n, a first pilot symbol of the next frame n 258 whilst producing corresponding displacement of four OFDM symbols with respect to the next pilot sub-carrier symbol 258 only produces a displacement of two OFDM symbols with respect to the last scattered pilot symbol 251 of the previous frame n-i. Thus as will be appreciated if passing the samples of the channel produced by the pilot sub-carrier symbols 251 and 258 through an interpolator or extrapolator filter a change in the relative displacement of the samples of the channel provided by the last pilot symbol 251 of frame n-I and the first pilot symbol 258 of frame n will cause an interruption or phase discontinuity of the sampling of the channel.

A better appreciation of the problem of the phase discontinuity explained above can be understood from an example of a conventional time interpolator 214 shown in Figure 6a.

First Example Interpolator One method of perfonning time-interpolation is called FIR-interpolation, in this method a conventional FIR interpolator is used to interpolate between scattered pilots in the time direction. Such an interpolator is shown as a block diagram in Figure 6a. In Figure 6a an example of a time interpolator 214.1, which corresponds in operation to the time interpolator 214 shown in Figure 4 is shown in more detail. As shown in Figure 4 the divider circuit 210 compares the received pilot symbols with a version of the pilot symbols as transmitted at the transmitter to generate samples of the channel on an input 280.1 to the time interpolator 214.1. The samples of the channel H/k are received at a zero-inserter 281, which increases the sample rate of its input by some factor (in this example Dy) by inserting Dy-i zero samples in between each input sample. The up-sampled version of the channel samples H/k are then received at a first input of a sequence of delay storage units 286.1 which form a shift register of a finite impulse response filter, which includes weight combining units 288.1 and an adder 290.1. In accordance with. a convention operation of a finite impulse response filter (FIR), a sample value currently stored in the delay storage unit 286.1 is weighted by a coefficient V71 of the impulse response of the filter and combined by the adder 290.1. The output of the adder circuit 290.1 generates for each of the samples 1, k of the channel a sample of the channel estimate H/k. If the frequency response of the FIR filter is chosen correctly, then its output will be an up-sampled version of the input. In a practical implementation, the zero-inserter and FIR filter could be combined to provide a more computationally efficient implementation.

Figure 6b shows an example of a time interpolator 214.1 which includes a controller unit 320.1 and a re-sample generator 324.1, which operate to compensate for a change in a relative phase of the samples of the channel provided by the pilot symbols, as a result of a change in the rate at which pilot symbols distributed in the OFDM symbols. The operation of this example of the present technique is explained as follows: Suppose the input samples x 400 of the zero-inserter block are numbered XM (with M from 0 to infinity) as shown in Figure 7a. Then outputs of this block yM*DY will be equal to input xDy 400 and all other outputs (yM*Dy+I... y*Dy+DyJ) 402 will be zero as shown in Figure 7b. Suppose the FIR filter has N taps numbered from T0 to TN . The output z of the FIR is the sum: z1 It can be seen that because yMDy±1 YM*Dy+nyi 402 will be zero as shown in Figure 7c, there is no point in calculating the contribution to z for these inputs because it is known that this contribution will be zero. Therefore, there is no point in inputting these zero-valued samples 402 into the input shift register. In this way, the number of calculations required to generated each output z is reduced by a factor of roughly Dy (depending on whether the number of taps in the filter is a multiple of Dy or not) and also the number of stages needed in the shift register is reduced by a similar factor.

Accordingly it will be appreciated that Figure óa and Figure 6b should be considered as illustrative only in that a typical implementation would combine the up-sampling tO with the The calculation of the output z 404 then goes through a cycle of Dy numbered Co to CDYi: C0 -shift the new sample into the shift register and calculate z using taps T0, To3, and all shift register values; Ci -calculate z using taps TDi, T2D..i... and all but the oldest or all of the shift register values; C0.1 -calculate z using taps T1, T0+1.. and all but the oldest or all of the shift register values.

The taps of FIR filters are usually symmetrical: that is tap T0 is the same as tap TN1; tap Ti = tap TN.2 and so on. This gives the FIR filter a linear phase response. If the taps are symmetrical, then the delay through the filter is (N-l)12. When FIR interpolation is used for time interpolation in a DVB-T2 receiver, then a compensating delay of (N-l)12 must be.put in the data path to compensate for the delay through the FIR interpolator so that data can be corrected by dividing by the correct channel estimate, as shown in Figure 4.

In DVB-T2 and other OFDM systems, the channel is sampled by scattered pilots which occur every Dy symbols. If these samples are passed through the system shown in Figure 6a, then the output will be up-sampled by a factor of Dy. In effect, the system shown in Figures 6a has interpolated between the scattered pilots 400' in order to generate intermediate channel estimates 402' as shown in Figure 7c. It can be seen that as the filter goes through the cycle Cc, to C0yj interpolated channel estimates 402' are generated for the OFDM symbols between the symbols that contain scattered pilots. If N is chosen to be odd, then because the delay through the FIR filter is (N- 1)12, the output of the filter at symbol S will be a channel estimate for symbol S-(N- 1)12. This is shown in Figure 8a. In Figure 8a in this example, Dy is 4, N is 25, the number of stages in the FIR shift regi.ster is 7. The FIR interpolator outputs a channel estimate for symbol 12 at symbol 24 using taps T0, T4, T8, T12, T16, T20 and T24. At that time, the 7 stages in the FIR shift register contain samples 406 of Symbols 0, 4, 8, 12, 16, 20 and 24. Figure 8b shows the situation one symbol later. It is because the symbol does not contain a scattered pilot on this carrier 408, that no new sample is input to the FIR shift register and it does not shift. Therefore, it still contains the samples 406 of Symbols 0, 4, 8, 12, 16, 20 and 24. I-Iowever, because taps T3, T7, T11, T15, T19 and T23 are used on this cycle to calculate the output is a channel estimate 410 for symbol 13 and only the six newest shift register locations are used; the oldest shift register location is unused. The situations for the next three cycles are shown in Figures 8c tO 8e. It can be seen that, because Dy is 4 in this example, the situation in 8e is similar to that in Figure 8a.

An example of the cycle Co to CDyI is shown in Figures 9a to 9e for an FIR interpolator with 17 taps where Dy is 4. In each Figure, the vertical lines represent OFDM symbols. A dotted vertical line 412 represents the first symbol of the next T2FRAME. A dotted vertical line 414 represents the last received OFDM symbol. The circles 416, 418, 420, 433 represent pilots. The rectangles represent the five locations 430, 432 in the shift register. The circle 420 marks the UFDM symbol for which the channel estimate is calculated by the FIR. interpolator. The circles 416, 418, 420 represent pilots that are used to calculate the channel estimate; the circle 433 represents a pilot that is not used to calculate the channel estimate. Figure 9a shows cycle C0: all five shift register locations 430 are used together with taps T0, Li, T5, T12 and T15. As shown in Figures 9a and 9e, the delay of the channel estimate calculated is delayed by (14-1)12 and so corresponds to the OFDM symbol stored in the central location of the shift register 420. Figure 9b shows cycle C1: only the newest four shift register locations 430 are used together with taps T3, T7, T11 and Tic. The oldest location in the shift register 432 is not used in the calculation. Again, the delay of the channel estimate calculated is delayed by (N-1)12 and so corresponds to the OFDM symbol 420. Figure 9c shows cycle C2: again the newest four shift register locations 430 are used together with taps T2, T6, T,c, and T14 and again the delay of the channel estimate calculated is delayed by (N-l)/2 and so corresponds to the OFDM symbol shown 420. Figure 9d shows cycle C3: the newest four shift register locations 430 are used together with taps T1, T5, T9 and T13 and again the delay of the channel estimate calculated is shown 420. Figure 9e shows the next cycle C11: now all five shift register locations 430 are used again together with taps T0, T4, T3, Tn and T16 and the delay of the channel estimate calculated again corresponds to the OFDM symbol stored in the central location of the shift register.

In DVB-T2 the regular spacing of the scattered pilots may be disturbed, for example at the boundary between two T2FRAMEs. The FIR interpolation method described here relies on the samples of the channel being taken at strictly regular intervals. If the intervals are regular, but with a sampling interval that changes occasionally, this will be manifested as a temporary error in the output channel estimate, which will last for N symbols, where N is the numbei of taps in the FIR filter. This situation is shown in Figures lOa to lOc. The FIR interpolation method operates correctly in Figure l0a to calculate the channel estimate 468 corresponding to OFDM symbol 8, because the spacing between the pilot-bearing symbols is regular (4). Likewise, the method also operates correctly when calculating the channel estimate corresponding to symbol 9, because again the spacing between the pilot-bearing symbols is regular. However, the spacing between the pilot-bearing symbols in Figure lOc is not regular: symbol 16 is pilot-bearing, as is symbol 18, so the gap is 2 whereas that between other pilot-bearing symbols is 4 and so the samples are not taken at regular intervals and the method will not operate correctly -there will be an error in the channel estimate for the next 17 OFDM symbols (because the number of taps in

this example is 17).

The example embodiment represented in Figure 6b operates to mitigate the situation shown in Figure lOc, where the spacing between the pilot-bearing symbols is not regular and the interval to the new sample 466 is smaller than the regular interval.

On an occasional sample point change, if the interval to the new sample 466 is smaller than the regular interval, then: 1) the new sample 466 is stored in the memory 2) the remainder of the samples in the memory are overwritten by new virtual samples, which are calculated by the re-sample generator 324.1 under th.e control of the controller unit 320.1 by interpolating between the original samples.

This example is illustrated in Figure 11, in which samples from a first T2FRAME 450 are shown on a first plot 452, with samples from a second T2FRAME 454, and in a second plot 456 there are shown virtual samples 458 calculated by interpolating between the original samples 450.

An example using this method is shown in Figures lOa to lOb for an FIR interpolator with 17 taps where Dy is 4. In each of Figures lOa to lOh, the vertical lines represent OFDM symbols. The doffed vertical line 412 represents the first symbol of the next T2FRAIVIE. The doffed vertical line 414 represents the last received OFDM symbol. The circles 464, 466, 468, 470, 474 represent pilots. The rectangles 430, 432 represent the five locations in the shift register. The circle 468 marks the OFDM symbol for which the channel estimate is calculated by the FIR interpolator. The circles 464, 468, 466, 470 represent pilots that are used to calculate the channel estimate; the circle 474 represents a pilot that is not used to calculate the channel estimate. As has already been described, the FIR interpolation method operates correctly in Figure 1 Oa to calculate the channel estimate corresponding to OFDM symbol 8, because the spacing between the pilot-bearing symbols is regular; the FIR interpolation method also operates correctly when calculating the channel estimate corresponding to symbol 9, because again the spacing between the pilot-bearing symbols is regular. Figure lOc shows the situation when the spacing between the pilot-bearing symbols is not regular; if nothing is done to mitigate this, there will be an error in the channel estimate for the next 17 OFDM symbols. Figure lOd shows the application of th.e above method. The new pilot sample is input into the newest shift register location (symbol 18) and then the remainder of the samples in the memory are overwritten by new virtual samples 470, 468, which are calculated by interpolating between the original samples. Figures lOe to lOh then show how the FIR interpolation method can work normally for subsequent OFDM symbols.

It can so happen that the gap between the inegularly spaced pilot bearing symbols is larger than the regular interval. In this case the re-sample generator 324.1 operates as follows: 1) the new sample is stored in the memory 2) the previously stored newest sample is overwritten by a new virtual sample, which is calculated by interpolating between the original newest sample and the new sample 3) the remainder of the samples in the memory are overwritten by new virtual samples, which are calculated by interpolating between the original samples.

This example is illustrated in Figure 12, in which the original samples 480 are shown with the first new sample 482 of the next T2FRAME, a virtual sample 486 calculated by interpolating between the original newest sample 482 and the virtual samples 484 calculated by interpolating between the original samples 480.

However, following the above method would result in there being a delay or gap in the output, because the system has to wait for the next irregularly spaced sample. In some systems (for example a system that is not operating in real time) this delay could be tolerated. In DVB-T2, however, such a delay could not be tolerated, so some other method is required to mitigate the situation where the gap between the irregularly spaced pilot bearing symbols is larger than the regular interval.

In DVB-T2, each T2FRAME starts with one or more P2 symbols. One feature of these P2 symbols is that there is a pilot on every possible pilot-bearing carrier. So, in a DVB-T2 system, there will always be a suitably positioned pilot in the intervening P2 symbol. However, the gap between the last pilot of the first frame and the pilot in the P2 symbol may not be the same as the regular spacing (Dy) and the gap between the pilot in the P2 symbol and the first pilot of the second frame also may not be the same as the regular spacing (Dy). So, the method described previously can be used twice to mitigate the situation where the gap between the irregularly spaced pilot bearing symbols is larger than the regular interval: once between the last pilot of the first frame and the pilot in the P2 symbol and again between the pilot in the P2 symbol and the first pilot of the second frame.

An example of using the method described previously twice to mitigate the situation where the gap between the irregularly spaced pilot bearing symbols is larger than the regular interval using an intervening P2 pilot is shown in Figures l3a to l3m again for an FIR interpolator with 17 taps where Dy is 4. Again, the FIR interpolation method performed by the resample generator 324.1 operates correctly in Figure 13a to calculate the channel estimate corresponding to OFDM symbol 8, because the spacing between the pilot-bearing symbols is regular; the FIR interpolation method also operates correctly when calculating the channel estimate corresponding to symbol 9, because again the spacing between the pilot-bearing symbols is regular. Figure l3e shows the situation when the last received OFDM symbol is a P2 symbol: the gap between the last pilot-bearing symbol and the P2 symbol is not the same is the regular spacing. Figure l3d shows how the previously described method can be used to mitigate this so that the P2 pilot 490 can be used to mitigate the fact that the gap between the last pilot of the first T2FRAME and the first pilot of the second T2FRAPvIF is greater than Dy. The P2 pilot sample 490 is input into the newest shift register location symbol 18 and then the remainder of the samples in the memory are overwritten by new virtual samples 492, which are calculated by interpolating between the original samples 464. Figures 13e and l3f show bow the FIR interpolation method works normally for the next two OFDM symbols. Figure l3g shows the situation when the first pilot of the second T2 frame becomes the last received OFDM symbol: the spacing between the previously sampled P2 pilot 490 and this first pilot of the second T2FRAME is not Dy. Figure l3h shows the application of the previously described method for a second time. The new pilot sample is input into the newest shift register location (symbol 21) and then the remainder of the samples in the memory are overwritten again by new virtual samples 492, 468, which are calculated by interpolating between the samples previously stored. Figures 13i to l3m then show how the FIR interpolation method can work normally for subsequent OFDM symbols.

Second Example Interpolator in Figure 14 a second example of a possible time interpolator 214 shown in Figure 4a is shown. As shown in Figure 4a the divider circuit 210 compares the received pilot symbols with a version of the pilot symbols as transmitted at the transmitter to generate samples of the channel on an input 280 to the time interpolator 2i4. The samples of the channel H are received at the first input 280 of an adder unit 282 which receives on a second input an output from a decaying integrator 284 which generates a mean value of the pilot estimate for scaling purposes. The output of the adder circuit 282 provides a filtered sample of the channel Gik which feeds a first input of a sequence of delay storage units 322 which form a shift register which corresponds in architecture to a fmite impulse response filter and forms with weight combining units 288 and an adder 290 a extrapolator or extrapolation filter. At the output of each of the storage units 322, a sample value currently stored in the delay storage unit 322 is fed to a weight combining unit 288 which multiplies the value in the storage unit by a scaling factor A result of applying the weighting factors ox to the samples of the channel estimate applied by the weighting units 288 is fed to an adder circuit 290. 1 tins tile output ot tile aocier circuit 290 generates tor eacn or th.e samples 1, k of the channel a extrapolated sample of the channel estimate Hik.

The adder unit 290 also receives the mean pilot estimate to value Mik which offsets the resulting interpolator sample value. The output of the adder unit 290 is fed to a further adder unit 292 which is combined with the most recently received sample of the channel FT, from the input 280 to form at an output of the adder circuit 292 an error value Cik on an output channel 294. The error value elk is formed between the extrapolated sample of the channel lk and the most recently received sample of the channel H. The error value can be used to provide a quality metric of the channel estimate and also provides an indication of a rate of change of the channel. The samples of the channel estimate formed by the extrapolator at the output of the adder circuit 290 feeds a linear interpolator 296, which includes a delay circuit 286 that provides a linear interpolation between the samples of the channel estimate in the time domain. Thus at the output of the linear interpolator 296 is formed an estimate of the channel which is fed on an output channel 300 to the frequency interpolator as shown in Figure 4a.

As will be appreciated effectively the extrapolator 289 forming part of the time interpolator 214 shown in Figure 14 generates the extrapolated samples of the channel estimate from previously received pilot symbols which is combined with interpolating between the extrapolated samples at each of the locations of the pilot symbols to produce the estimate of the channel for equalisation. However in order to accommodate the change in the rate of arrival of the samples of the channel, a control unit 320.2 is provided which controls adapted storage delay storage units 322.

Furthermore a pilot sample adjusting unit 324.2 is also controlled by the controller 320.2. The controller 320.2 has a control input 326 which controls delay units 328 as will be explained shortly. The output of each of the delay unit 328 feeds a scaling unit 330 which receive on a second input an output from the previous corresponding stage of the delay unit 322. Thus the delay storage units 328 within the pilot adjuster 324.2 correspond relatively to the delay storage unit 322 provided within the extrapolation filter 214.

In operation the controller 320.2 receives an input 121 from the control unit 119 which provides an indication of the location of the pilot symbols within each of the OFDM symbols which are being received. Thus at any point the controller 320.2 is provided with an indication of a relative sample position from the OFDM symbols of the received pilot symbols. The controller 320.2 in one example is therefore provided with am' indication of the pilot pattern which is currently being used by DVB transmitter.

In operation the controller 320.2 detects when a phase discontinuity occurs and controls the delay storage units 322 of the extrapolator 289 and the delay storage units 328 to shift the content of the output of the delay storage units 322 of the main interpolator into the delay storage units 328 via connecting channels 334. The content of each of the delay storage unit 322 of the extrapolator 289 are therefore shifted and stored into the corresponding delay storage units 328. The multiplying units 330 then weight the content of the delay storage units 328 by a sample corresponding to the input of the corresponding delay storage unit 322 to form at an output a combination of the two values weighted by a weighting factor y.

The pilot adjusting unit 324.2 provides one example for accommodating a change of phase in the relative position of the samples of the channel provided by the pilot subcarrier pattern.

In one example the change in the pilot symbol position, which changes the position of the resulting sample of the channel is accommodated by simply passing the new sample of the channel provided by the newly received pilot symbol location through the extrapolator 289. However this can take more than j times the spacing between the samples of the channel provided by the pilot symbols (Dy), where j is the number of taps in the extrapolator, because time is required for the taps to be adjusted correctly, in order for the extrapolator 289 to recover the correct estimate of the channel which can produce a degradation in performance parlicularly for example in dynamic channels. This conventional situation is illustrated in a graphical representation shown in Figure 15 which corresponds to the example shown in Figure 5. As shown in Figure 15 treating the first sample of the channel provided by the pilot symbol at the position 258 as if the pilots were in phase is effectively shifting the pilot signal samples from the previous frame n-i backwards by two positions to equivalent new positions 360. However, although the pilot samples are then in phase with the new position of the pilot sample 258 from frame n a result of interpolation between the pilot signal samples at their new position 360 is to generate a extrapolated equivalent value of the new pilot symbol 362. As can be seen from a difference between the extrapolated estimate of the channel sample 362 and the actual channel sample 258 there is an error 364.

As explained above an interpolator in accordance with the present technique includes the pilot adjusting unit 324.2 as shown in Figure 14 which is arranged to generate adjusted values of the samples of the channel to the effect that the new sample of the channel provided by a pilot symbol for example from the next frame which arrives at an interval which is different to the interval for which the samples have previously been received can be passed into the interpolator reducing the error in the interpolated samples of the channel estimate. In particular the pilot adjusting unit 324.2 generates new samples of the channel by adjusting the samples which are presently stored within the delay storage units 322 so that these correspond in time to a sampling interval which is in phase with the new sample. Thus the controller 320.2 controls the delay storage units 328 and 322 in order to shift out the samples from the extrapolator filter which are present in the delay storage units 322 and re-calculate these samples by interpolating between these samples to generate new samples which provide a sampling interval which is in phase with the new sample produced from the pilot symbol at a sub-carrier location which would otherwise be out of phase with the old pilot symbols. Thus according to the present technique, once the controller 320.2 detects that the last in-phase sample before a discontinuity in the sampling interval between the samples of the channel has been reached, the controller controls the delay storage units 322 to copy the content of the delay storage units 322 into the S corresponding delay storage units 328 within the pilot adjusting unit 324.2. Thereafter the controller 320.2 arranges for the output of the scaling units 330 to be clocked back into the delay storage units 322 within the time interpolator 214 after the combining units 313 have multiplied and scaled the contents of the delay storage units 328 with the input of the corresponding delay storage unit 322 to form effectively an interpolation between the two sample values. Effectively therefore the operation of the pilot adjusting unit 324.2 shown in Figure 14 performs the operation shown in Figure 16 in which new samples of the channel 380 are interpolated between the sample values 251 corresponding to the location of the pilot sub-carriers for the previous frame n-i. Thus as shown in Figure 16 five new samples of the channel are generated 380 from the existing six samples 251 within the extrapolator delay storage units 322 and one new sample of the channel is generated 385 from the newest sample of the channel previously stored within the extrapolator delay storage units 322 and the last in-phase sample before a discontinuity in the sampling interval between the samples of the channel has been reached 255. These samples are generated and provided by the pilot adjusting unit 324.2.

Further Adaptations As will be appreciated from the explanation of an example embodiment described above, the present invention provides a pilot adjusting unit which adapts the samples of the channel generated by pilot symbols so that these samples are in phase with a sample of the channel generated by a newly received pilot symbol which would otherwise be out of phasc with the pilot symbols used to generate the samples of the channel until that point. As will be appreciated any other difference between a relative position or an interval between previous samples and a newly received sample may be accommodated with appropriate adaptations being made to the structure of the pilot adjusting unit 324.2. In some examples a distance between the pilots at a point of discontinuity may be greater than the distance between sampling intervals produced by the pilot symbols until that point. For example if a distance between samples of the channel is equal to Dy, which is a separation between a position of one scattered pilot and a number of OFDM symbols before that scattered pilot appears again at that position in an OFDM symbol, then the extrapolator can only extrapolate Dy symbols ahead. In Figure l7a an example is shown in which the distance between the pilots at a point of discontinuity may be greater than the distance between sampling intervals produced by the pilot symbols until that point. In order to function correctly, the extrapolator requires two inputs: a new sample of the channel and an error value e1 which is the difference between the extrapolated channel estimate and the actual channel sample. For the example shown in Figure l7a an error value C/k produced between the extrapolated sample and the new sample is false because a location of the sample corresponds to the position of a data bearing sub-carrier rather than a pilot sub-carrier. Because the error value C/k 15 false and the sample of the channel 1420 is not a pilot sub-carrier, some way of mitigating this situation is required.

Two options for dealing with this situation are illustrated in Figure I 7b and l7c. A solution for generating a extrapolated plot value from which an error sample can be generated is provided in Figure l7b. For the example shown in Figure 17b a P2 pilot is used to replace the sample of the channel which is within Dy sub-carrier symbols of the previous sample of the channel and therefore can be used to generate interpolated sample values using the pilot adjusting unit 324.2. Thus at a location where a pilot is available within Dy symbols of the latest sample 1422, a pilot symbol is used from a P2 symbol which is within Dy symbols of the last scattered pilot symbol 1422 at a location 1424. Thereafter the pilot adjusting unit 324.2 can adjust the value of the samples of the channel provided within the delay storage units 322 as explained above. Now, the error value e produced between the extrapolated sample and die new sample is false because the location of the sample input to the extrapolator does not correspond to the position its output. To mitigate this situation, an error value C/k of zero can be used. Then the process described previously can then be repeated to adjust the extrapolator for the change in sample timing between thc P2 pilot 1424 and the first scattered pilot of the new frame 1426, which will be within Dy sub-carrier symbols of the P2 pilot 1424.

An alternative example solution is provided in Figure 1 7c. In Figure l7c the extrapolator 289 is used to generate a sample of the channel at position 1430 with respect to the previous sample of the channel 1432. The process described previously can then be used to adjust the extrapolator for the change in sample timing between sample 1430 and sample 1436 which can be performed by the pilot adjusting unit 324.2. However, the error value elk produced between the extrapolated sample 1430 and the new sample is false because the location of the sample corresponds to the position of a data bearing sub-carrier rather than a pilot sub-carrier and the new sample cannot be input to the extrapolator because it is not a sample of a pilot sub-carrier. To mitigate this situation, the extrapolator output can be input to the extrapolator in place of the new sample and an error value e, of zero can be used (because the extrapolator output is used as its input, the error value elk will indeed be zero). In this way the extrapolator car' continue operating correctly and can go on to extrapolate further carriers such as 1436.

Summary of Operation

ln summary an example operation of the present technique is summarised in Figure 18 with each of the steps of the method being formed to address a technical problem of aligning the samples values of the channel estimate when there is a change of the relative spacing between the pilot symbol sub-carriers produced by the OFDM symbols. The steps shown in Figure 18 are explained as follows: is illustrated in Figure 18, which is summarised as follows: SI: As a first step it is necessary to detect whether there will be a change in a relative sampling of the channel as a result of a change in the pilot pattern, either as a result of the rate of occurrence of the pilot symbol sub-carriers or the position of the sub-carriers.

52: After detecting the change in the relative sampling of the channel, a revised relative position of the samples of the channel is determined from the pilot symbol sub-carriers which have already been received in order to produce a sampling rate and position of the samples of the channel which is the same as the position of the new pilot symbol sub-carrier. The determined revised relative position therefore corrects out of phase samples present in the extrapolator memory so that they are in phase and can be used with the sample of the channel provided by the new pilot sub-carrier symbol.

54: Once the position of the samples of the channel has been determined, the sample values are then re-calculated for the channel in accordance with the revised relative position of the samples so that these are in phase with the samples provided by the new pilot symbol sub-carrier.

56: The re-calculated sample values are then provided to an appropriate extrapolator/interpolator.

58: The re-calculated samples of the channel estimate are combined with the new sample of the channel estimate provided by the newly received pilot symbol.

Various modifications may be made to the embodiments herein before described. For example it will be understood that the particular component parts of which the channel estimator, extrapolator and interpolator described are logical designations. Accordingly, the fimetionality of these component parts may be manifested in ways that do no conform precisely to the forms described above and shown in the diagrams. For example aspects of the invention may be implemented in the form of a computer program product comprising instructions that may be implemented on a processor stored on a data sub-carrier such as a floppy disk, optical disk, hard disk, PROM, RAM, flash memory or any combination of these or other storage media, or transmitted via data signals on a network such as an Ethernet, a wireless network, the Internet, or any combination of these of other networks, or realised in hardware as an ASIC (application specific integrated circuit) or an FPGA (field programmable gate array) or other configurable or bespoke circuit suitable to use in adapting the conventional equivalent device.

Embodiments of the present invention may also fmd application with other appropriate transmission standards such as the terrestrial DVB transmission standard known as DVB-C2 or the hand-held standard known as DVB-NGH at the time of writing. For the example of DVB-C2, it will be appreciated that the OFDM symbols are transmitted and received via a cable so an appropriate adaptation of the transmitter and receiver architecture can be made. However, it will be appreciated that the present invention is not limited to application with DVB and may be extended to other standards for transmission or reception, both fixed and mobile.

Although illustrated with an example interpolator which includes a extrapolator with memory elements which contain samples of a channel estimate generated from previously received pilot symbols, other example embodiments may provide interpolators which do not use a extrapolator, but some other arrangement which use previously received samples of the channel. Furthermore other differences in sample spacing may occur other than the ones illustrated in the explanation of the present technique presented above.

Claims

CLAIMS1. A receiver for receiving data from a sequence of OFDM symbols transmitted via a channel, each of the OFDM symbols comprising a plurality of data bearing sub-carriers on which the data is transmitted and a plurality of pilot symbol bearing sub-carriers on which pilot symbols are transmitted, the pilot sub-carriers being arranged within the OFDM symbols in accordance with a predetermined pilot sub-carrier pattern, the receiver including a channel equaliser comprising a pilot symbol extractor for extracting the received pilot symbols from the pilot sub-carriers of each of the received OFDM symbols, a channel estimator which is arranged in operation to form, for each pilot sub-carrier, an estimate of a sample of the channel through wtiich the received OFDM symbols have passed by comparing the received pilot symbols with a version of the pilot symbols transmitted wiLh the OFDM symbols, and to interpolate the samples in time and/or frequency to produce an estimate of the channel, and an equaliser, which is arranged in operation to reduce the effects of the channel on the received OFDM symbols using the estimate of the channel generated by the channel estiniator, so that data can be recovered from the received OPDM symbols, wherein the channel estimator includes an interpolator which is adapted to detect a change in a relative sampling of the channel derived from a change in an interval between the pilot symbol sub-carriers which have been received and a newly received pilot sub-carrier symbol, to determine an adapted relative position of the samples of the channel provided by the pilot symbol sub-carriers which have been received to produce a relative position of thc samples of the channel which corresponds to a sampling position for the newly received pilot symbol sub-carrier, to re-calculate the samples of the channel estimate in accordance with the adapted relative position of the samples, arid to use a sample of the channel estimate derived from the newly received pilot symbol sub-carrier in combination with the re-calculated samples of the channel estimate to generate the interpolated samples of the channel estimate.
2. A receiver as claimed in claim 1, wherein the interpolator is configured an interpolator which is adapted to detect that the relative sampling position provided by the pilot symbols changes from a regular spacing of original sampled to a spacing for a new sample which exceeds the regular spacing of the original samples, to store the new sample in the memory, to interpolate between an original newest sample and a new sample, which has been previously written into memory of the interpolator, tO to overwrite the remainder of the samples in the memory by new samples, which are calculated by interpolating between the original samples.
3. A receiver as claimed in claim 1, wherein the interpolator is configured an interpolator which is adapted to detect that the relative sampling position provided by the pilot symbols changes from a regular spacing of original samples to a spacing for a new sample which is less than the regular spacing of the original samples, to store the new sample in memory of the interpolator, and to over-write the remainder of the samples in the memory of the interpolator by new samples, which are calculated by interpolating between the original samples.
4. A receiver as claimed in Claim 2 or 3, wherein the interpolator memory includes a plurality of delay storage elements which are arranged sequentially to receive the samples of the channel estimate derived from the pilot symbol sub-carriers and the interpolator includes a combiner which combines the samples from the delay storage elements with weighting factors to form for the newly received pilot symbol sample, a sample of the channel estimate and a controller, the controller being arranged in operation to detect the change in the relative sampling of the channel derived from the newly received pilot sub-carrier symbol, to determine the adapted relative position of the samples of the channel estimate provided by the pilot symbol sub-carriers in accordance with the relative position of the newly received pilot symbol sub-carrier with respect to the pilot symbol sub-carriers which have been received, to retrieve the samples of the channel estimate from the storage delay elements and to re-calculate the samples of the channel estimate in accordance with the adapted relative position of the samples, and to generate a sample of the channel estimate from the storage delay elements with the re-calculated samples of the channel estimate for interpolating to form the channel estimate.
5. A receiver as claimed in Claim 4, wherein the interpolator includes a pilot adjusting unit, the pilot adjusting unit including a plurality of secondary delay storage elements, the pilot adjusting unit being arranged in operation under control of the controller to receive the samples of the channel estimate stored in the delay storage elements of the interpolator memory, to store the samples of the channel estimate in the secondary delay storage elements of the pilot adjusting unit, to interpolate between the samples of the channel estimate stored in the secondary delay storage, to replace the samples of the channel estimate in the delay storage elements with the re-calculated samples, to generate a sample for the new position of the samples of the channel resulting from the newly received pilot symbol, and to interpolate between the samples.
6. A receiver as claimed in any preceding claim, wherein the change in the relative sampling position caused by the newly received pilot symbol is as a result of a change in the pilot pattern.
7. A receiver as claimed in Claim 6, wherein the change in the pilot pattern occurs from one frame to the next frame.
8. A receiver as claimed in any preceding Claim, wherein the OFDM symbols are transmitted in accordance with a Digital Video Broadcast standard.
9. A method of receiving data from a sequence of OFDM symbols transmitted via a channel, each of the OFDM symbols comprising a plurality of data bearing sub-carriers on which the data is transmitted and a plurality of pilot symbol bearing sub-carriers on which pilot symbols are transmitted, the pilot sub-carriers being arranged within the OFDM symbols in accordance with a predetermined pilot sub-carrier pattern, the method comprising extracting pilot symbols received from the pilot sub-carriers of each of the received OFDM symbols, forming, for each pilot sub-carrier, an estimate of a sample of the channel through which the received OFDM symbols have passed by comparing the received pilot symbols with a version of the pilot symbols transmitted with the OFDM symbols, interpolating the samples in time and frequency to produce an estimate of the channel, and reducing the effects of the channel on the received OFDM symbols using the estimate of the channel, so that data can be recovered from the received OFDM symbols, wherein the interpolating includes detecting a change in a relative sampling of the channel derived from a change in an interval between the pilot symbol sub-carriers which have been received and a newly received pilot sub-carrier symbol, determining an adapted relative position of the samples of the channel provided by the pilot symbol sub-carriers which have been received to produce a relative position of the samples of the channel which corresponds a sampling position for the newly received pilot symbol sub-carrier, re-calculating the sample values of the channel estimate in accordance with the adapted relative position of the samples, and using the sample of the channel estimate derived from the newly received pilot symbol sub-carrier in combination with the re-calculated samples of the channel estimate to generate the interpolated samples of the channel estimate.
10. A method as claimed in claim 9, wherein the detecting a change in a relative sampling of the channel includes detecting that the relative sampling position provided by the pilot symbols changes from a regular spacing of original sampled to a spacing for a new sample which exceeds the regular spacing of the original samples, and the re-calculating the sample values and the using the sample of the channel estimate derived from the newly received pilot symbol sub-carrier in combination with tbe re-calculated sample values to generate the interpolated samples of the channel estimate, includes storing the new sample in memory, interpolating between an original newest sample and a new sample, which has been previously written into the memory of the interpolator, and overwriting the remainder of the samples in the memory by new samples, which are calculated by interpolating between the original samples.
11. A method as claimed in claim 9, wherein the detecting a change in a relative sampling of the channel includes detecting that the relative sampling position provided by the pilot symbols changes from a regular spacing of original samples to a spacing for a new sample which is less than the regular spacing of the original samples, and the re-calculating the sample values and the using the sample of the channel estimate derived from the newly received pilot symbol sub-carrier in combination with the re-calculated sample values to generate the interpolated samples of the channel estimate, includes storing the new sample in memory of the interpolator, and over-writing the remainder of the samples in the memory of the interpolator by new samples, which are calculated by interpolating between the original samples.
12. A method as claimed in Claim 9, 10 or 11, wherein the interpolating between the samples of the channel estimate formed from the received pilot symbols includes providing a plurality of delay storage elements, arranging for the samples of the channel to be received sequentially by a plurality of delay storage elements, and combining the samples of the channel estimate from the delay storage elements with weighting factors to form for the newly received sample of the channel estimate, wherein the re-calculating the sample values of the channel estimate in accordance with the adapted relative position of the samples of the channel estimate includes retrieving the samples of the channel estimate from the delay storage elements, re-calculating the samples of the channel estimate in accordance with the adapted relative position of the samples, and re-loading the storage delay elements with the re-calculated samples of the channel estimate for use in generating an extrapolated sample of the channel estimate.
13. A method as claimed in any of Claims 9 to 12, the method including providing a pilot adjusting unit, the pilot adjusting unit including a plurality of secondary delay storage elements, and the re-calculating the samples of the channel estimate includes receiving the samples of the channel estimate stored in the delay storage elements of the extrapolator, storing the samples of the channel estimate in the secondary delay storage elements of the pilot adjusting unit, interpolating between the samples of the channel estimate stored in the secondary delay storage, replacing the samples of the channel estimate in the extrapolator delay storage elements with the re-calculated samples, generating an extrapolated sample for the new position of the samples of the channel resulting from the newly received pilot symbol, and interpolating between the extrapolated samples.
14. A method as claimed in any of claims 9 to 13, wherein the change in the relative sampling position caused by the newly received pilot symbol is as a result of a change in the pilot pattern.
15. A method as claimed in Claim 14, wherein the change in the pilot pattern occurs from one frame to the next frame.
16. A method as claimed in any of Claims 9 to 15, wherein the OFDM symbols are transmitted in accordance with a Digital Video Broadcast standard.
17. A computer program providing computer executable instructions which when loaded onto a computer causes the computer to perform the method according to anyofClaims9to 17.
18. A data sub-carrier having a recording medium, the recording medium having recorded thereon a computer program according to Claim 17.
19. A receiver as substantially as herein before described with reference to the accompanying drawings.
20. A method of receiving substantially as herein before described with reference to the accompanying drawings.