SU1550516A1 - Method of determining procedure of message transmission - Google Patents

Abstract

The invention relates to computing, in particular, is intended for use in local area networks with bus topology for controlling the transmission of data packets through a common channel. The purpose of the invention is to reduce the time to resolve conflict situations. The method of determining the message transmission order is that at each station transmitting, the duration of its transmission from its beginning to the termination of transmission in case of collision detection is measured, the interval from the beginning of the first collision to the end of the second collision or to the beginning of the first pause, the interval from the beginning of the second collision or from the end of the first bubble to the end of the third collision or the beginning of the second pause, etc., compare the duration of its transmission with other measured intervals and based on the results obtained set the order of transfer. 3 il.

Description

The invention relates to computing and is intended for use in local computer networks with bus topology for controlling the transmission of data packets through a common channel.

The purpose of the invention is to reduce the time to resolve conflict situations.

FIG. 1 shows a state and transition diagram; in fig. 2 - time diagrams of the processes of collision transmission; in fig. 3 is a timing diagram of network operation after collision of transmissions.

FIG. Figure 1 shows a state and transition diagram for this method and the following notation is used: SEC = initial state; AO - zero active state; A1 is the first active state; A2 is the second active state; PD - transmission.

FIG. 2 shows time diagrams of the collision processes of two (a) and three (b, c, d) transmissions. Time intervals are shown on them: TD - the first measured interval, transmission duration A; T g- is the second measurable interval, transmission duration B; TC is the third measured interval, transmission duration C; Tr - pause between gears B and C; T% - the wrong time interval, if

ate

ate about

Oi

0

A station cannot distinguish between B & C transmissions; tq, is the interval of forced Clash.

FIG. 3 - shows a timing diagram of the network after a collision of three gears. It is marked: AHVHC collision of three gears (without detail); A, B, C - successful transmissions of stations A, B and C; X - successful transmission of the new active Station X; tp; - pause between gears (i 1,2,3); TP2 - pause between Gears C and X; Txd is the waiting time for access to the common channel of stations X.

In a local area network using Multiple jna with carrier control and Collision Detection (MDCN / OS), all stations are connected to one common channel through which data packets are transmitted, at the beginning of which the destination address is contained. Stations continuously listen to a common channel and, when a data packet appears in it, check its destination address. If the destination address of the packet coincides with the address of this station, then the latter accepts (this packet. With these explanations in mind, consider the essence of the proposed method.

All passive stations (which do not have a data packet ready for transmission) are in the initial state. They Jra6o expect a data packet from their Upper levels, or they form it themselves. They also listen to a common K channel, determine its state (Free, Successful Transmission, Collision) and, if necessary, will degrade their addressed data packets. When a station-ready data packet appears in the station, this station enters state A2 (transition 1 in FIG. 1).

Stations located in the states of ISH and A2 form time intervals Tn 2 at each disappearance of hens, in the common channel. For these stations, the common channel is considered free not immediately, but after the expiration of the aforementioned interval Tn2, if during its formation no carrier of new transmission appeared. If during the Formation of Tp2 a carrier of new transmission appears, then the formation of Tp2 stops, and the common channel for this station is considered to be continuously occupied. In the prototype TP2 Tpo 9.6 µs. it

five

0

five

0

five

0

five

0

the minimum allowable pause between transmissions in the common channel is necessary so that the end of one transmission does not merge with the beginning of the other. In the proposed method, typically t5 6 Tp2 2ts.

The station, having entered the active state A2, waits for the release of the common channel (if it is occupied) and immediately switches to the AT state (transition 2), i.e. starts the transmission (if the common channel is free). In a PD state, there is usually one station and it successfully transmits its data packet. As usual, the station checks its transmission and if it goes successfully, it continues until the transmission of the entire data packet and then returns to the initial state (transition 3). Thus, with a small load, more precisely, in the absence of collisions, the operations of the data transmission control process in the LAN do not differ from the prototype (except for the value of Tn2, which in this case is not significant).

If in the PD state there are at least two stations, then each of them will detect other people's transmissions (i.e., will detect a collision of transmissions). The station then fixes the transmission (i.e., continues for a given tm, for example, 3.2 µs at a transmission rate of 10 Mbps, i.e., four bytes or 32 bits) and then stops its transmission (as in prototype).

In contrast to the prototype, transmission control consists in the fact that at a station participating in a collision, during the collision process, the time intervals Ta and Tb (with m 2) or Ta, T and Tc (with m 3), etc. are measured to obtain the information necessary to control the transmission.

Although FIG. 2 shows cases of m 2 and m 3, the proposed method is not limited to these examples. Four or more intervals are measured in the same way in the case of th 4. In addition, instead of Td and Tj- in all cases, it is possible to measure and use further, for example, Td Tq - tq, and / or

five

T T - (Fig. 2a), if this is more convenient in practical implementation, since the value of tq is small and fixed and its addition or subtraction is not significant,

After measuring the duration of all transmissions of a single collision process, i.e. after the end of the collision process, the stations participating in this collision which are in the PD state start forming two time intervals - Tpo and, where Tpo - the minimum pause between transmissions, a Tnl t5 (for example, 5.2 x as in the prototype). At the same time, at each of these stations, the results of their measurements are compared and a queue is established, for example, in the order of decreasing values of the measured intervals. For example, for the process of FIG. 2b, a Tq Tc Tg is obtained, thus each station has determined its place in the queue. The first is station A (since Ta Tg, Tc), the second is C (since Ta4tc t8) and a third is B (since Tn, TL TV). The queue can be set up with ТЈ) and in increasing order, i.e.

T g- Tt - Td for the same example.

Q

but, until all stations in order go from Ai to AO and then to OP, i.e. until all the participants in the collision have transmitted their data packets.

Consider the operation of the LAN using the proposed transmission control method and under the condition of Tn2 2tg. As already mentioned, with a small load, as long as there are no collisions, the operation of the LAN occurs as in the prototype, so we start with a collision of, for example, three transmissions ВхВхС (Fig. 3). After measuring the intervals T, TЈ and Te by comparing them in size, all three stations determine their place in the common queue and transmit their data packets in order.

If a new active station X appears at that time, it goes to state A2 and waits for the release of the common channel. But since t „Тп2 (i 1,2,3), the common channel for it is occupied until all the participants

0

This merely changes the order of the collision dock; it will not transfer its packets to the queue, which has no meaning. The main thing is that there should be one order common to all stations.

After determining its place in the queue, the station, having received the first place, switches to the AC state, and all other (not the first) stations switch to the A1 state (transition 6). In the AO, the station continues to form the Tpo interval, and in A1, the Tp1 interval. After the TpO expires, the data station from the AO enters the PD state, i.e. starts the transfer. If, for the reasons discussed below, the first transmission does not start, after the Tnl expires, the station from A1 will go to the AT state, i.e. begin transmission. In the case of m 2, such a station is only one, so it successfully transmits its packet. In the case of m e 3, a second collision will follow.

If the first transmission starts and passes normally, then every time, as a station

in A1 it detects a successful transmission in jo of the first collision,

in the common channel, they increase their place in the queue (the second station becomes the first, the third - the second, etc.), and the new first station changes to the AO state. After disappearing in the common carrier channel, the next first station in the AO again forms the Tpo, and the remaining stations (in A1) form the Tnl, etc., as already stated. Possible values of 2t are acceptable and Tn2S but most preferable from t5 Tp2 6 Thus, for different installations, you can set the current value of Tn2, or the best filling

0

you. Then, Tp2 2ts appears in the common channel, indicating that all the collision participants have already transmitted their packets, and therefore station X enters the AT state (starts its transmission).

Now we introduce the condition Tn2 ts. If all stations correctly identified their places in the queue, then the LAN operation is no different from what has already been stated, only Tp2 pause is shorter. But for reasons that are discussed in more detail below, it may happen that some place in the queue is left empty. Then the corresponding transmission begins and a pause appears in the common channel. As already mentioned, as soon as this pause becomes ts Tnl, the stations from A1 begin to transmit. In this case, simultaneously with them, the transmission starts and X, a collision occurs, etc., but station X is now involved, do not wait until all the participants have completed the transfer

five

0

first collision

Possible values of Tn2 are not limited to 2t3 and t5, since Tp2 C is also acceptable and Tp2 2 ts, but the most preferable are t5 Tp2 6 2ts. Thus, for different operating conditions, you can set the most appropriate value of Tn2, providing either the best filling of the common channel

gears, or a smaller average time | waiting for access Thd „

Finally, we consider some cases in the work of a LAN.

If two stations begin transmitting strictly at the same time, then Ta Tg may appear, as in the case shown in FIG. 2a Then the first two (or two second) stations may appear in the queue. This will lead to a repeated collision, new dimensions, etc. Then, because of the different equipment and subscriber lines, etc. already after one or two tfki repetitive collisions will have - with # TQ / Tg- and two successful transfers will follow. Another solution is also possible. As soon as the stations detect the equality Tq Tg1, they go into the idle state and stop trying to transmit this packet.

Another special case is shown on f | 2g, when at the entrance of station A the end of transmission B coincides with the beginning of transmission C. Then station A cannot distinguish between Tg- and Tc and measures one common interval of 1%. As a result, either the first two or two second station. In addition, some place in the queue may be empty.

Thus, the considered and other possible special cases do not disrupt the work of the LAN. In addition, all are unlikely and therefore can be very rare.

As the load increases, the number m of collision participants also increases. Accordingly, the number of measurable time intervals also increases, and the likelihood of their equality and other special cases also increases, which could impair the work of the LFV. This can be easily eliminated, for example, if m M, where M is the maximum number of collision participants with which the equipment can cope, then all m participants of this collision are reset to their original state as well as the prototype after the specified maximum number

five

0

five

0

five

0

five

0

(16) failed transmission attempts. In other words, part of the input load is reset, for which the rest is serviced efficiently and effectively, i.e. possible degradation of LAN performance during overloading.

In order to implement the proposed method, each station must have standard (known, for example, in a prototype) means for determining the state of a common channel (free, successful transmission, collision); for forming, storing, receiving and transmitting data packets; to encode, decode, synchronize and verify the transmitted data bits; and others, such as those that exist or may be on standard or other LANs using MDT / OC.

In addition, equipment is needed to implement the newly introduced measurements of measuring the time intervals Ta, Tg-, Tc ..., and to compare the numbers obtained, as well as to store their sequence number in the transmission queue.

Claims (1)

Invention Formula The method of determining the order of transmission of messages in the computer network, is that at each station of the computer network control the channel and start the transfer, if the channel is free, characterized in that, in order to reduce the time to resolve conflicts, each station has transmission, the duration of its transmission from its beginning to the transmission termination in case of collision detection, the interval from the start of the first collision to the end of the second collision, or until the beginning of the first pause , The interval from the beginning of the second end of collision or from a first space to the end of the third collision, or before the second pause, etc., the duration of its transmission is compared with other measured intervals and on the basis of the results set transmission sequence. at tp - t „ figure 1 1ШШ  -r - 3 nz Fm.g t