RU2209521C1 - Exchange method for channel-access dedicated-server network - Google Patents

Abstract

FIELD: computer engineering; communication networks; appointed- channel access division methods. SUBSTANCE: common method designed for data exchange in dedicated-server networks provides for automatic transfer of service functions to any other network subscriber in case of server failure by transmitting broadcast pulse chain to channel during origination of each time segment of channel band and choice of subscriber whom server functions will be transferred by counting number of time segments within which broadcast synchronizing pulses were not received; subscriber is switched over to server mode through definite time interval counted off from the moment of reception of last broadcast synchronizing pulse chain. EFFECT: enhanced reliability of network and effective use of entire band of channel. 1 cl, 1 dwg

Description

 The invention relates to the field of computer systems, communication networks, mainly to deterministic methods for sharing access to multiplex channels.

 The known interface is a serial single-wire digital information control system (GOST 26765.52-87) that regulates the presence in the information channel of a device that controls the transfer of data between this device and another device, and between other devices. GOST 26765.52-87 specifies control commands, data transmission and physical parameters of signals. Each device can operate in one of three modes: controller, terminal device), monitor. The exchange in the channel is controlled by a device in controller mode. Only one device per channel can be in controller mode. The types of commands issued by the controller are divided into ten formats. Formats with a number from 1 to 6 inclusive are addressable, and each subscriber who receives such a package is required to give a response word. Commands from 7 to 10 format are multicast (broadcast). When receiving such commands, the terminal device should not give a response word. A device operating in monitor mode cannot transmit response words and any data.

 The disadvantage of GOST 26765.52-87 is the use of the monitor mode to provide the backup functions of the controller, which requires the use of an additional device in the network that is not involved in the exchange until the failure of the subscriber performing the functions of the controller.

 The well-known "Protocol for the transmission of packet data for radio communications" according to the patent of the Russian Federation 2139636 (H 04 L 29/06, H 04 L 12/56), in which a dedicated control station is used to control the exchange.

 The disadvantage of this protocol is the low reliability of the network due to the inability to transfer network management functions to any subscriber.

 The prototype of the proposed method is "a method of exchanging data between a plurality of subscriber stations over a wireless LAN through a central control station" according to the application of the Russian Federation 97109431 (H 04 L 12/00).

 The disadvantage of this method is the low reliability of the network due to the inability to transfer network management functions to any subscriber. If the main station fails, several subscribers can take control functions at the same time, but the claimed method does not provide for actions that ensure the selection of one of several subscribers.

 The objective of the invention is to create a single method of exchange in networks with a dedicated service node, increasing network reliability and ensuring the efficient use of the entire channel band for various network configurations and network flows.

где Тз - время обслуживания одной заявки;
То - среднее время опроса абонентов;
выбирают период цикла обслуживания Тц из условия: Тц>(То+Тз•Ам)/2; обслуживают каждого абонента до достижение (равенства) количества обслуженных заявок вычисленной ранее величине 'Nз', после чего переходят к обслуживанию другого абонента; производят удаление из списка обслуживания, абонентов, не ответивших несколько раз подряд (как минимум два) и перемещают в список доступных всех возможных абонентов с периодом большим, чем 'Тц', как минимум в два раза и минимально необходимым для требуемой задержки восстановления связи с абонентом.An exchange method is proposed, consisting in the fact that the channel strip is divided into a predetermined number of time segments, at the beginning of each time segment, a broadcast synchronizing message is sent from a dedicated service node (hereinafter controller), which contains its own address 'Ak', while the addresses of subscribers, including the controller, choose in the range [O..Am], where 'Am' the maximum possible number of subscribers served, and the transition of the subscriber with the address 'A' to the controller mode is carried out in the absence of broadcast x synchronization packages through the time interval 'Tn', which is counted from the moment of receipt of the last broadcast synchronization package and calculated by the formula
Tn = Tk + K • Tc • (A-Ak), if Ak <A;
or Tn = Tk + K • Tc • (Am + A-Ak), if Ak> A,
where Tk - the duration of the transition to controller mode;
K is the number of service cycles during which there was no reception of synchronizing packages;
TC - the duration of the service cycle period;
Am - the maximum value of the subscriber's address;
interrogate available subscribers and determine the total number of responses containing service requests; determine the number of 'From' applications from each subscriber processed in one service cycle according to the formula
Nз = 1+ (integer part)

where T3 is the service time of one application;
That is the average polling time of subscribers;
choose the period of the service cycle TC from the condition: TC> (T0 + T3 • Am) / 2; serving each subscriber until reaching (equality) in the number of served requests to the previously calculated value of 'Nz', after which they proceed to service another subscriber; remove from the list of services subscribers who have not answered several times in a row (at least two) and move to the list of available all possible subscribers with a period greater than 'TC', at least twice and minimally necessary for the required delay in reconnecting with the subscriber .

The implementation of this method is reduced to performing certain actions on the part of the program with respect to the interface device of a serial single-wire digital information control system. The program is executed on a dedicated service node, before starting a cyclic survey of all available network subscribers with a period of TC, generates a command word of format 10 (a group control command with a data word, according to GOST 26765.52-87) and a data word that contains its own controller address in the memory of the serial interface device 'Ak.' The addresses of subscribers, including the controller, are selected in the range [O..Am], where Am is the maximum possible number of subscribers served. Next, a command is issued to send a message in format 10, as a result of which the device issues a broadcast synchronizing message to the bus. On the side of the terminal device, each such package is received and the internal timer is synchronized (reset) at the time of this reception. The need to switch to controller mode for the subscriber with the address 'A' operating in the terminal device mode occurs when there are no broadcast synchronizing packages after the time interval 'Tp', counted from the moment of receiving the last broadcast synchronizing package and calculated by the formula:
Tn = Tk + K • Tc • (A-Ak), if Ak <A;
Tn = Tk + K • Tc • (Am + A-Ak), if Ak> A,
where Tk - the duration of the transition to controller mode;
K - the number of service cycles during which there was no reception of synchronizing packages;
TC - the duration of the service cycle.

 The drawing shows a timing diagram of the transfer of functions of the controller to the subscriber with the address An> Ak.

где Тз - максимальное время обслуживания одной заявки;
То - среднее время опроса.A further sequence of program actions on the controller side is as follows. In the device’s memory, a request chain in format 5 is generated for all available subscribers. A command is issued to execute the first request in the chain. Next, the query chain is automatically executed, as a result of which data words received from terminal devices are stored in the device memory. If the received word contains the sign of the transfer request, then this request is stored in the table of service requests. After analysis of all received data, a survey of subscribers takes place, requests from which were placed in a table. If a subscriber does not answer twice in a row, then it is marked as unavailable. After receiving all the answers and determining the total number of Nc answers containing service requests, the controller determines the number of Nc requests from each subscriber, which he will be able to process in one service cycle with the duration of the TC period by the formula
Nз = 1+ (integer part)

where T3 is the maximum service time for one application;
That is the average polling time.

 According to this formula, in the case Nc = Am, the integer part will be equal to zero and, therefore, the controller will need to process only one application in a cycle. The period TC is selected from the condition TC> (T0 + T3 • Am) / 2, which provides a balance between the average service delay in the case of a full network load and in the case of a minimum load. The service of each subscriber continues until (the equality) of the number of served requests to the previously calculated value of N3, after which the controller proceeds to service another subscriber. Independence of the service algorithm from the number of available subscribers and maximization of bandwidth are ensured by removing from the service list subscribers who did not answer twice in a row, and by moving to the list of available all possible subscribers with a period greater than TC, at least twice and minimum necessary for the required delay in restoring communication with the subscriber.

 Thus, the proposed method of organizing the exchange provides automatic transfer of controller functions to subscribers who are in terminal device mode, with automatic selection of one of the subscribers who are ready to switch to controller mode. At the same time, for all remaining subscribers, the possibility of exchanging information with the subscriber who has switched to controller mode remains.

 Thanks to the appearance of the above capabilities, the network reliability is significantly increased, since a complete network failure occurs only when one subscriber remains. In all other cases, the failure of any of the subscribers does not lead to disruption of the exchange between other subscribers, whereas in the prototype the failure of the central station, i.e. just one failure, would lead to a complete network failure. In addition, the proposed exchange method guarantees the full use of channel resources, regardless of the number of subscribers and their current needs for data transmission on the channel.

 As a consequence of the above functional characteristics, this exchange method can be used without any modifications in a system with a variable number of subscribers, while ensuring efficient use of the communication channel.

The claimed technical solution is not obvious, and its technical level is quite high. The possibility of industrial application of the invention is not in doubt, since
- the method is intended for use in a specific field, namely in computer technology, in particular in control and management systems.

- the implementation of the method is possible using the operations described in the application;
- the method allows to achieve the claimed technical effect.

Claims (6)

1. The method of exchanging information between subscribers for a network with a dedicated channel access service node, which consists in the fact that the channel strip is divided into time segments - K, characterized in that at the beginning of each time segment from the dedicated service node - the controller is transmitted to the channel broadcast a synchronization package, which includes the controller’s own address - Ak, while the addresses of subscribers, including the controller, are selected in the range from 0 to the maximum possible number of served subscribers - Am and o there is a choice of a subscriber who is in terminal device mode, to which controller functions will be transferred by counting by each serviced subscriber the number of time segments K, duration Tc, during which there was no reception of broadcast synchronizing packages and transfer the subscriber with address A to controller mode after a time interval Tn, which counted from the moment of receiving the last broadcast synchronizing package and calculated by the formula
Tn = Tk + K • Tc • (A - Ak), if Ak <A;
Tn = Tk + K • Tc • (Am + A - Ak), if Ak> A,
where Tk - the duration of the transition to controller mode;
K is the number of time segments during which there was no reception of broadcast synchronizing packages;
TC - the duration of the time segment;
Ak is the controller’s own address;
Am - the maximum value of the subscriber’s address,
and they also survey available subscribers and serve subscribers based on responses containing applications received during a survey of subscribers.  2. The method according to claim 1, characterized in that after issuing a broadcast synchronizing message and polling available subscribers, the total number of responses containing service requests is determined. 3. The method according to claim 1, characterized in that the number of requests N3 from each subscriber, which can be served in one time segment with a duration of TC, is determined by the formula

where T3 is the service time of one application;
That is the average polling time of available subscribers;
Nc is the total number of responses containing service requests. 4. The method according to claim 3, characterized in that the duration of the time segment TC is selected from the condition
Tc> (T0 + T3 - Am) / 2.  5. The method according to claim 3, characterized in that each subscriber is served until the number of served requests is reached (equal) to the previously calculated value of N3, after which they proceed to service another subscriber.  6. The method according to claim 1, characterized in that they remove from the list of service subscribers who have not answered several times in a row (at least two), and move to the list of available all possible subscribers with a period greater than TC at least twice and minimum necessary for the required delay in reconnecting with the subscriber.