GB2452316A - Computer resource management unit that selects an optimiser for a resource based on the operating conditions of the computer - Google Patents

Abstract

Disclosed is a method for managing the device or communication resources of a computer. The method involves the resource management unit allocating a resource to a requesting entity based on set performance constraints and selecting an optimiser for the resource based on performance criteria, operational status and/or situation. The management unit runs the optimiser and is able to select an alternative optimiser if the performance of the selected optimiser diverges from a desired performance level. The management unit may rerun the selection process if the performance diverges over a set threshold, and the process may be iterative. The computer may have performance data gathering means that collects data about the performance of the apparatus executing the entity, the operational status of the computer and performance constraints imposed by the entity. The operational status data includes power availability, radio signal strength, communication means, memory capacity, processor and computer location. The performance constraints include security level, communication speed, quality of service and data compression rate.

Description

1 2452316 Computer Apparatus and Method of Controlling the Same The present invention is concerned with the control of computer hardware using software, taking account of resources and other factors which might affect performance.

Aspects of the invention are particularly concerned with the optimisation of application performance and resource utilisation on a communications terminal. More specifically, aspects of the invention are concerned with the provision of an adaptive system for dynamically configuring and scheduling communications resources and application behaviour using a situation-aware strategy.

Modern computerised communication devices are multipurpose and are equipped with powerful processors and multiple communications interfaces. This allows a user to direct a communications device to run numerous communicating applications simultaneously. The applications in question will include not only user applications but also protocol, control and other support mechanisms that contend for networking resources and for data and control information transmissions. One of the main challenges in the design and implementation of this type of device is to provide Quality of Service for multiple applications, with potentially conflicting requirements, which are competing for resources. This challenge is especially true for wireless communications devices.

Computing devices, required to run several applications simultaneously, adopt numerous strategies to manage and share the available communications resources among the applications. Such management and sharing of available resources can involve optimisation techniques.

A number of techniques described in prior art documents explore various options for resource optimisation of communication systems.

The majority of these solutions utilise inter-layer communications allowing an entity (that is, a protocol or algorithm) belonging to one layer of the OSI model to interact with an entity belonging to another, allowing monitoring and controlling of several layers at the same time. "Cross-layer design: a survey and the road ahead" (Srivastava, V.; Motani, M.; Communications Magazine, IEEE Volume 43, Issue 12, Dec. 2005 Pages: 112 -119) describes several examples of a cross-layer approach.

The literature on resource and protocol optimisation can be broadly classified into two groups, namely frameworks and algorithms.

By the term "frameworks", the present discussion refers to resource management and cross-layer frameworks. Such frameworks are methods for information access, information storage and control of various configurations at terminal and network level, as described in "A Tutorial on Cross-Layer Optimization in Wireless Networks" (Lin, X.; Shroff, N.B.; Srikant, R.; Selected Areas in Communications, IEEE Journal on Volume 24, Issue 8, Aug. 2006 pages:1452 -1463) and the paper by Srivastava et al. noted above. These frameworks often provide flexible interfaces which are able to abstract user/application requirements and resource capabilities, and which enable the development of optimisation or management algorithms.

Recent research on reconfigurable software architectures often implies use of such frameworks in the optimisation process. For example, use of user profiles based on explicit user feedback and/or pre-defined user/application requirements (preferences) to personalise services has been studied in the domain of reconfigurable systems, for instance in "Mode switching and QoS issues in software radio" (A.H. Aghvami, T.H.

Le, N. Olaziregi IEEE Personal Communications 8:55, 38-44, 2001).

US71 11297, US71371 19 and W006052565A3 each describe a centralised resource management framework for wireless communication devices, which utilises an entity, called a resource manager, to access and control the available resources through resource providers. US7I 11297 in particular describes a method allowing dynamic resource configurations by mapping application specified configurations and provider-offered configurations. This framework also describes the use of a policy manager to set the resource allocation policies using the policy rules stored in a database.

US2005044205 describes a framework to enable resource management on a computing device by incorporating multiple communications options as well as other application parameters in the decision making process.

Each of these approaches describes a resource management framework for computing devices in general. However, the particular requirements involved in the implementation of a wireless communications device are overlooked.

On the other hand, much of the literature on the technical implementation of resource optimisation and management algorithms aims for the optimisation of the performance of one specific application (or application class such as multimedia). An example of this approach is disclosed in "Application-driven cross-layer optimization for video streaming over wireless networks (Khan, S.; Peng, Y.; Steinbach, E.; Sgroi, M.; Kellerer, W.; Communications Magazine, IEEE Volume 44, Issue 1, Jan. 2006 pages 122-130).

Some algorithms achieve this by adapting the application behaviour to match available link resources, whilst others focus on optimising the performance of lower layer aspects such as the MAC or link layer of a single wireless technology (usually WLAN) to match application requirements (for example "A Cross-Layer Framework for Wireless LAN QoS Support": Giovanni Pau, Daniela Maniezzo, Shirshanka Das, Yujin Lim, Janghyuk Pyon, Heeyeol Yu, Mario Gerla,, IEEE 2003). Both of these approaches involve the monitoring and controlling of application requirements and communications resources.

The field of the present invention can be considered as extending at least to implementations of frameworks or algorithms in achieving management of resources in a wireless communications apparatus.

An aspect of the invention provides a computer apparatus, e.g. a wireless communications apparatus, operable to provide a device resource or communications resource which is consumed by one or more resource consuming entities, said computer apparatus including resource management means operable to determine, on the basis of one or more performance constraints and/or operation status or situation, an allocation of said resource to said one or more resource consuming entities, said resource management means being operable to select a resource optimiser in view of said one or more performance criteria, said resource management means being further operable to run said resource optimiser selected by said optimisation selection means and yet further operable to reselect an alternative resource optimiser if performance of said selected resource optimiser diverges from a desired performance level in respect of said one or more performance constraints.

The computer apparatus may store a plurality of optimisers, each optimiser being appropriately operable for a respective set of performance parameters.

The computer apparatus may be operable to seek an optimiser stored in a remote location.

The resource management means may be operable to determine that said performance diverges from said one or more performance constraints on the basis of one or more performance thresholds.

The resource management means may be operable to monitor performance of said selected resource optimiser on an iterative basis.

The computer apparatus may include performance information gathering means operable to collect information relating to the performance of said apparatus executing said one or more entities.

The performance information gathering means may be operable to collect information regarding the operational status of the computer apparatus.

The operational status may include at least one of local consumable power availability, power source, received communications signal strength, available communications mode, available memory capacity, available memory type, available processor speed and physical location of the computer apparatus.

The performance information gathering means may be operable to collect information regarding one or more performance constraints imposed by said one or more entities.

The performance constraint information may include one or more of security level status, required communications speed, quality of service, and data compression rate.

Another aspect of the invention provides a method of managing a processing resource of a computer apparatus, said processing resource being available for processing one or more processing entities in said computer apparatus, said method comprising selecting an optimiser in accordance with one or more performance constraints and running said optimiser to allocate said processing resource, as available, to said one or more processing entities, said method further comprising monitoring, during said running, for a change in condition of said computer apparatus with regard to said one or more performance constraints and, on detection of a change in condition, repeating said step of selecting an optimiser.

The method may include storing a plurality of optimisers, each optimiser being appropriately operable for a respective set of performance parameters.

The method may include seeking an optimiser stored in a remote location.

The method may include determining if said performance diverges from said one or more performance constraints on the basis of one or more performance thresholds.

The method may include monitoring performance of said selected resource optimiser on an iterative basis.

The method may include collecting information relating to the performance of said apparatus executing said one or more entities.

Collecting may include collecting information regarding the operational status of the computer apparatus.

Collecting may include collecting information regarding one or more performance constraints imposed by said one or more entities.

One aspect of the invention involves an implementation which operates on the basis that no single optimisation can suit every scenario. In particular, this aspect of the invention comprises a method for realising meta-optimjsation based solutions, which exploits various types of performance optimisers dealing with different parts of the system or even different performance criteria.

One aspect of the invention provides an adaptive resource optimisation method intended to configure a communications device by dynamically selecting optimisers with a strategy based on operational status.

Another aspect of the invention provides an adaptive resource optimisation method for computer communications devices that allows selection of a suitable optimisation strategy dynamically.

In such a method, plug-in based optimisers may be used, each of which can be inserted or removed in real-time, providing flexible and adaptive resource management.

A situation-aware strategy may be employed, to select the most suitable optimiser at any given instance. Situation detection of wireless devices may be defined as the process in which the device establishes its "operational status" or current state within its operational environment.

Cross layer methods may be used, to derive device situation by monitoring dynamics of application requirements, application priorities, resource availability, and quality of communications paths. Device situation could encompass computing, user, physical or time context and could relate to different system or operational aspects of the device.

A resource supervisor may be provided, to authenticate registration and to authorise actions performed by plug-in resource optimisers.

The plug-in resource modules may use a standard (or common) application progranmling interface to access and control the communications resources of the computing device.

The abovementjoned situation aware strategy may employ terminal wide communications resource information and application requirements stored in storage means. Such storage means may be maintained in a central manner, or may be distributed e.g. locally or remotely.

The plug-in resource optimiser may use the dynamics of terminal wide application requirements. Such an aspect of the invention may be operable such that it is only triggered when any of such attributes are changed, and the change crosses a predefined threshold.

Aspects of the invention so specified may be employed to access the application requirements explicitly registered by the application in the storage means.

An embodiment of the invention may be used to derive application requirements implicitly based on notification conditions registered by an application.

Dynamic registration of new applications and application requirement updates may be enabled by aspects of the invention in real-time.

Dynamic registration of new communications resources (i.e. links and devices) and resource updates may be enabled in real-time.

Privilege or access rights associated with the applications may be used to assign priorities to all active applications in run-time.

Aspects of the invention may be operable to configure all concerned links and communications devices of a communications network in real-time with minimum disruption to ongoing communications. Aspects of the invention may perform optimisation at a terminal level or may perform network wide resource optimisation.

An aspect of the invention may be able to implement the above without the need for incorporation thereof completely into an operating system or network stack.

Another aspect of the invention provides management of an available device resource or communications resource of a computer apparatus which is available for use by one or more resource consuming entities embodied in the computer apparatus and comprises selecting an optimiser in accordance with one or more performance constraints and running the optimiser to allocate the resource, as available, to the one or more entities.

This aspect further comprises monitoring, during running, for a change in condition of the computer apparatus with regard to the one or more performance constraints and, on detection of a change in condition, repeating the step of selecting an optimiser.

Aspects of the invention may comprise either originally implemented processing instructions or hardware implementation in a computer (for instance a communications enabled computer) or may comprise processor instructions which can be introduced thereafter. The processor instructions can be included in a computer program product which could be embodied in a storage medium, or a hardware device such as a mass storage device, a removable ROM or similar device, or could be delivered from a remote location by way of a signal bearing computer receivable information which can then be determined, for instance by at least one of decoding, decompression, unbundling or similar steps, into a series of program instructions for execution as an aspect of the invention in the computer apparatus.

Further aspects and advantages of the invention will become clear to the reader on consideration of the following description of specific embodiments of the invention, with reference to the accompanying drawings, in which: Figure 1 is a schematic diagram of an architecture in accordance with a conventional approach to resource management in an computerised wireless communication device; Figure 2 is a schematic diagram of a wireless communications apparatus of the invention; Figure 3 is a schematic diagram of an architecture established in the wireless communications device illustrated in Figure 2; Figure 4 is a first implementation of a method of resource optimisation in accordance with the specific embodiment; and Figure 5 is a second implementation of the method of resource optimisation.

As illustrated in Figure 1, system architectures adopted by the prior art can be summarised as comprising a generic operating system and protocol stack 10 onto which applications 20 can be installed. A first application 20 is illustrated as including an optimisation tool 22, and interacts with the operating system and protocol stack 10 by virtue of an API 24. Further applications 20 can be included ("plugged in") as illustrated.

The operating system and protocol stack 10 supports interactions between the applications and network resources, which are presented to the operating system by virtue of network interfaces 30, each of which can include an optimisation tool 32. An API 34 offers the operating system and protocol stack 10 with the facilities provided by the network interface 30.

A resource manager 40 is provided which includes algorithms and facilities to establish, as far as possible, the matching of available resources to the demands placed by the applications to be offered to the user.

Figure 2 discloses a wireless communications device 100 in accordance with a specific embodiment of the invention.

The communications device 100 comprises a processor 120 operable to execute machine code instructions stored in a working memory 124 and/or retrievable from a mass storage device 122. By means of a general-purpose bus 130, user operable input devices 136 (if supplied) are in communication with the processor 120. The user operable input devices 136 comprise any means by which an input action can be interpreted and converted into data signals, for example a keypad or a speech recognition unit.

Audio/video output devices 138 are further connected to the general purpose bus 130, for the output of information to a user. Audio/video output devices 138 include any device capable of presenting information to a user, for example, a screen on which text or graphical information can be presented to the user.

A communications unit 132 is connected to the general-purpose bus 130, and further connected to an antenna 150. By means of the communications unit 132 and said antenna(s) 150, the device 100 is capable of establishing wireless communication with other wireless devices within its coverage area, or with a base station.

In the device 100 of Figure 2, the working memory 124 stores applications 126 which, when executed by the processor 120, cause the establishment of an interface to enable communication of data to and from wireless stations. The applications 126 thus establish general purpose or specific computer implemented utilities and facilities that are used in providing the user with the requested facilities, using the available processing and transmission resources.

Figure 3 illustrates structurally the architecture implemented as part of the present embodiment of the invention. As noted above, the device 100 offers a plurality of network interfaces 200, two of which are illustrated as being in situ and one being illustrated as being introduced at the time of illustration. These can be, for example, a Bluetooth compatible driver and a software defined GPRS modem.

An API 210 is provided for the purpose of collecting, in the device, information concerning network facilities, their performance and their control. The API is able to refer to a database 220 which stores data, as will be described in due course. As before, an operating system (OS) and protocol stack 230 is provided, alongside a generic link management module 232.

In the layer above these two entities, an API 240 is provided, which implement handlers for events, queries and commands. This provides a common layer interposing between applications 250 installed in the device and the network hardware and other facilities which are required to implement the applications.

A resource optimisation unit 300 is further provided. This includes a situation detector 310, a plug-in resource manager 320, and a resource supervisor 330. As a whole, the resource optimisation unit 300 integrates with the application API 240 to enable the plugging in of specific optimisation algorithms to change the behaviour of the architecture of the system according to changes in the communications context.

To do this, the situation detector 310 provides interaction, via the API 240, with information regarding the state of the various facilities of the architecture which effect efficient delivery of application functions. The situation detector 310 has a global view of the overall system, acquired through continuously monitoring the dynamics of the system resources, the system users and other environmental factors. It makes use of the illustrated cross layer framework, and a common API (e.g. GOLLUM API: T. Famham, A. Gefflaut, A. Ibling, P. Mähönen, D. Melpignano, J. Riihijärvi and M. Sooriyabandara -Toward Open and Unified Link-Layer API Proceedings of the 1ST Mobile and Wireless Summit 2005, Dresden, Germany, June 2005) to access the resource availability and application status information collated in the shared database 220.

The resource supervisor 330 has overall control of the resource optimisation unit 300.

The unit includes the facility to find available optimisations, the selection of suitable optimisers, activating or plugging in selected optimisations, and controlling the mechanism that analyses the perceived performance improvement.

The plug-in resource manager 320 is but an example of a plurality of resource managers which will be available for use by the device, to implement performance optimisation algorithms aimed at improving one or more system aspects or performance parameters.

Such entities can be developed either by network operators, device manufacturers or independent software vendors. This depends upon the pmducer of such an entity adhering to the plug-in method described herein.

It will be appreciated by the reader that the described method is but an example, and does not dictate any specific implementation type. For instance, each of these components could be implemented either as user space entities, kernel space entities or even as a combination of those two.

The operation of the resource optimisation unit 300 will now be described further with regard to Figure 4, which reflects a first implementation thereof. In step Si -2, the situation encountered and experienced by the architecture is monitored. For the purpose of this disclosure, it will be understood that the term "situation" is a generic term designed to encompass one or many different criteria which are considered by the designer of a specific system to be important to the operation of an architecture in accordance with the best possible operating conditions.

For instance, the situation may include location related information. This could be geographical (such as provided by a GPS satellite), or relative (such as determining that the device is being used in an office or at home, in accordance with previously defined attributes or as informed by the user). It may also include possible measurements regarding radio environment, such as the identity of the access point (or indeed base station) to which the device has wireless communication contact.

The situation may also include information relating to the question of the use of applications in the architecture. A user may direct initiation or termination of various applications in the device. When a new application is initiated, or an existing application is terminated, this may be considered to result in a change of situation.

Further, there may be changes to policies defined by or for stakeholders in the use of the device, such as rules regarding access to data and/or applications. Security criteria may be changed. These, again, may represent changes in the situation.

The status of the hardware supporting the implementation of the architecture may also be included in the situation to be monitored in step SI -2. This may include power status information. Power status information may include whether a battery is being used, the remaining power stored in the battery, or whether an external power source (such as mains power) is being provided at the time.

In order to monitor the situation, the situation detector 310 gathers information required to make a decision on situation change. For instance, regarding the location of the device, this could be determined as a result of the device being GPS enabled, it could be as a result in the change of MAC address, or as a result of access point information changing.

In a simple example, the situation may be monitored, for appropriate changes, by the establishment of rules, e.g. thresholds on attributes and/or measurement parameters, which could be expressed in the SQL programming language. A suitable rule might be to detect when the signal is below a particular level. This may include also the use of a utility function, which will inevitably increase the complexity of the determination of whether the situation has changed, but may provide a more suitable implementation in certain situations. Instead of, or in combination with fuzzy logic, the rules could be Boolean, in making the final determination as to whether the situation has changed, as in step S 1-4. Finally, the step S1-4 in which a determination is made needs to involve a simple binary test but this may be supported by a fuzzy calculation of the extent to which certain underlying rules have been fulfilled.

If the situation changes, as determined in step Si -4, then the process proceeds to step S 1-6, where a selection is made of a suitable plug-in resource manager from storage means in the device. This is the purpose in keeping the resource supervisor 330 separate from the plug-in resource manager 320 as illustrated in Figure 3. It allows third party software vendors to determine solutions to various optimisation problems, which can then be implemented by way of a different resource supervisor 330. This increases flexibility. It will be appreciated that this is not essential to the implementation of the present examples of the invention.

Different plug-in resource managers 320 can offer different approaches to optimisation.

For instance, one plug-in resource manager 320 may be appropriate for channel selection (for instance in a case wherein received signal power is somewhat diminished), whereas another plug-in resource manager 320 could provide better facilities for power management if the hardware reports that available power is limited.

The question of selecting a suitable plug-in resource manager may not be a simple choice. This may in itself be a multi-criteria decision. The question may be as to how much memory resource should be expended on implementing a plug-in resource manager which will manage power more effectively, given that a high memory expenditure will itself give rise to a demand on power.

Moreover, certain constraints on this selection may override others. There may be a requirement for a security demand to be placed upon the device, which would override any plug-ins which made certain assumptions regarding the availability of data transmission characteristics which would in some way compromise security. The resource supervisor 330 making the selection step SI -6 would take this into account.

Should a new plug-in not be found then, in step SI -8, the process returns to the start.

This ensures that the existing implementation is not changed, regardless of any changes in the situation, if the present situation cannot be improved.

Should a new plug-in be found then, in step SI -10, the new plug-in resource manager 320 is inserted. This may have a consequence of removing an existing plug-in resource manager 320 already installed in situ.