JPH0478063B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (a) Technical Field of the Invention The present invention relates to a reception buffer management system, particularly in a packet switching network that allows a terminal to occupy a buffer for reception packets within a pre-registered number of buffers for reception packets. Regarding reception buffer management method.

(b) Prior art and problems Figure 1 is a diagram showing an example of a conventional reception buffer management method in this type of packet switching network.
FIG. 2 is a diagram showing an example of the packet transfer process in FIG. 1. In FIG. 1, a packet switching equipment 1 constituting a packet switching network controls two-to-two communication between a plurality of terminals accommodated therein. Packet switch 1
has a plurality of buffers 3 that temporarily store packets arriving at each terminal 2. Each terminal 2 determines the maximum number of buffers that each terminal 2 can occupy during communication (hereinafter referred to as the number of registered buffers m) based on the packet communication traffic that is expected to arrive simultaneously during communication, Register in packet switch 1. On the other hand, the packet switch 1 in which the number m of registered buffers has been registered by each terminal 2 sends the number of buffers occupied by each terminal 2 communicating at a certain point in time (hereinafter referred to as the number of occupied buffers q) to the buffer management unit 4. If the number q of occupied buffers exceeds the number m of registered buffers, a process (described later) is performed to deny reception of packets to the terminal 2. In FIG. 2, terminal 2B is terminal 2
In order to perform packet communication with A, when a call request packet CR is sent to the packet switch 1, the packet switch 1 sends the incoming packet CN to the terminal 2A.
Send. If the terminal 2A that received the incoming call packet CN is able to communicate, the incoming call acceptance packet CA
is sent back to packet exchange 1. The packet switch 1 that received the incoming call-allowed packet CA sends both terminals 2
A call is set up between A and 2B, and a connection completion packet CC is returned to terminal 2B. In the above process, the terminals 2A and 2B set the window size ws for the call, that is, the number of packets that can be continuously transmitted by the terminal 2B without waiting for the return of a signal indicating the receivable state from the terminal 2A. In the call shown in FIG. 2, it is assumed that the window size ws is set to 3. Thereafter, the terminal 2B sequentially transmits data packets DT (ps) to which transmission sequence numbers ps have been assigned. The terminal 2A that received the data packet DT (ps = 2) with the transmission sequence number ps = 2 sends a delivery confirmation and the data packet DT (ps = 2).
= 3) A sendable packet RR (pr = 3) is returned to which a reception sequence number pr = 3 is assigned to indicate that the packet is ready to receive the following packets. Packets that can be sent from terminal 2A via packet switch 1
The terminal 2B that received the RR (pr=3) sends the data packet DT (ps=3) to the data packet DT (ps=3) based on the preset window size ws=3.
=5) starts to be transmitted.

On the other hand, the packet switch 1 transmits data packets sent from each terminal 2 including the terminal 2B to the terminal 2A.
Every time DT (ps) is stored in the buffer 3, the buffer management unit 4 monitors the number of occupied buffers q, and stores the data packet DT sent by the terminal 2B.
(ps=4), the number of occupied buffers q has already been reached.
has reached the number of registered buffers m, packet switch 1 sends a congestion notification packet to terminal 2B.
The RNR is returned to temporarily stop the transmission of the data packet DT (ps), and the received data packet DT
(ps=4) is discarded.

As is clear from the above explanation, in the conventional reception buffer management system, the terminal 2B sends data packets DT based on the set window size ws.
However, because the number of occupied buffers q exceeds the number of registered buffers m, the transmitted data packet DT (ps) is discarded and must be retransmitted after the congestion is resolved, and the packet switch 1
However, the processing capacity must be consumed to discard the data packet DT and return the congestion notification packet RNR.

(c) Purpose of the Invention The purpose of the present invention is to eliminate the drawbacks of the conventional reception buffer management system as described above, and to realize a means for reducing invalid processing of packet switching equipment and terminals caused by congestion of the reception buffer. exist.

(d) Structure of the Invention The object of the present invention is to enable a packet receiving terminal to send a signal indicating that it is ready to receive packets in a packet switching network that allows a terminal to occupy a buffer for receiving packets within a preregistered number of buffers for receiving packets. The number of buffers determined by the window size set in advance at the time of transmission is added to the number of occupied buffers, and after confirming that the addition result does not exceed the registered number of buffers, the packet transmitting terminal sends a signal allowing packet transmission. This is achieved by providing a means for returning the

(e) Embodiment of the invention An embodiment of the invention will be described below with reference to the drawings. FIG. 3 is a diagram showing a reception buffer management system according to an embodiment of the present invention, and FIG. 4 is a diagram showing an example of the packet transfer process in FIG. 3. Note that the same reference numerals indicate the same objects throughout the figures.
In FIG. 3, various packets transmitted by a packet receiving terminal (terminal 2A in this example) are sent back to a packet transmitting terminal (terminal 2B in this example) via a buffer management section 40 provided in packet switch 1. be done. In FIG. 4, terminal 2B sends a call request packet CR to terminal 2A to packet switch 1 in the same way as in FIG. When the call acceptance packet CA is returned, the packet analysis circuit 41 in the buffer management unit 40 analyzes the returned incoming call acceptance packet CA, activates the buffer management circuit 43, and returns the incoming call acceptance packet CA.
It communicates that it is activated by. First, upon receiving the incoming call acceptance packet CA from the terminal 2A, the buffer management circuit 43 extracts and holds the number of occupied buffers q held by the packet exchange 1 corresponding to the terminal 2A, and also outputs the incoming call acceptance packet CA. , extracts the window size ws set for the call, and stores it. Next, the buffer management circuit 43 checks that the result of adding both the extracted occupied buffer number q and the window size ws does not exceed the registered buffer number m corresponding to the terminal 2A, and then adds the addition result (q+ws). The new number q of occupied buffers is notified to the packet exchange 1 and the number q of occupied buffers is updated (q+ws→q). Then, the buffer management circuit 43 updates the number of occupied buffers q by adding the window size ws in this way, and thereafter secures a buffer area in advance for the data packets DT (ps) that will be continuously transmitted from the terminal 2B. , instructs the packet sending circuit 42 to send an incoming call acceptance packet CA to the terminal 2B. Packet exchange 1 sets up a call between terminals 2A and 2B as in FIG. 2, and terminal 2B returns a connection completion packet CC. Thereafter, the packet switch 1 transmits the data packet DT to the terminal 2B.
(ps) to the terminal 2A, 1 is subtracted from the number of occupied buffers q held. data packet
Terminal 2A that received DT (ps) can send packets
When RR(pr) is returned, the packet analysis circuit 41
analyzes the returned ready-to-send packet RR(pr),
The buffer management circuit 43 is activated and the call identification and notification that the activation is based on the transmittable packet RR (pr) are transmitted. The buffer management circuit 43 first extracts the number q of occupied buffers and the window size ws held in the packet exchange 1 corresponding to the terminal 2A. Next, regarding the call, from terminal 2A to terminal 2
Latest sendable packet RR (pr) returned to B
is analyzed, and the received sequence number Pr is extracted and retained. Furthermore, the previous transmittable packet RR (pr)
Extract the reception sequence number pr' held when receiving. Thereafter, the following calculations are performed using the information (q, ws, pr, pr') retained by the above extraction.

That is, first, the number b of packets whose delivery has been newly confirmed by the terminal 2A this time is calculated from the confirmed sequence number pr held from the current transmittable packet RR (pr) and the packet whose delivery was already confirmed last time. It is determined from the difference from the sequence number ps'(pr-pr'→b). Next, the result b is tentatively added to the number q of occupied buffers, and the result q is compared with the number m of registered buffers, and it is determined whether the addition result q (=q+b) exceeds the number m of registered buffers. If the addition result does not exceed the number m of registered buffers, the number q of occupied buffers held by the packet switch 1 is updated based on the addition result, and the packet analysis circuit 41 updates the received sendable packet RR (pr). It is transferred to the sending circuit 42 and sent out. The packet switch 1 transmits packets that can be transmitted as in FIG.
RR(pr) is sent back to terminal 2B. If the addition result exceeds the number m of registered buffers, the packet sending circuit 42 is not sent the ready-to-send packet RR(pr), and the timing registration circuit 45 is made to respond to the call at the time point indicated by the timing circuit 44. and register it. Thereafter, when a preset time has elapsed, the selection circuit 46 activates the buffer management circuit 43 again to identify the call and send the sendable packet RR(pr).
It communicates that it is activated by. The buffer management circuit 43 determines the transmittable packets in the same process as described above.
Determine whether RR(pr) can be returned. That is, by waiting to send the ready-to-send packet RR (pr) until the transmission data packets DT (ps) from the packet exchange 1 to the terminal 2A are sequentially transmitted and the number of occupied buffers q decreases, the next data from the terminal 2B is transmitted. Rejects sending bucketed DT (ps=pr).
Furthermore, the communication ends and terminal 2B transfers to packet switch 1.
Terminal 2
When the terminal 2A that has received the disconnection instruction packet CI sends the disconnection instruction packet CI to terminal A and returns the disconnection confirmation packet CF, the packet analysis circuit 41 analyzes the disconnection confirmation packet CF and starts the buffer management circuit 43. , the call identification and disconnection confirmation packet CF conveys that the call has been activated. Here, among the window size ws that is added to the number of occupied packets q at the time of call connection and is kept constant during communication, it is added in advance to the number of occupied buffers q for the data packet DT from terminal 2B. A process of subtracting the buffer number b=pr+ws-ps-1 is performed (q-b→q). However, the latest values held are used for the reception sequence number pr and the transmission sequence number ps. In addition, the number of buffers occupied by data packets DT whose delivery has not yet been confirmed is determined by subtracting 1 from the number of occupied buffers q every time a data packet DT (ps) is transmitted from the packet exchange 1 to the terminal 2A. Since it disappears, the buffer management circuit 43 leaves it alone. Thereafter, the packet management circuit 43 instructs the packet sending circuit 42 to send out the disconnection confirmation packet CF received by the packet analysis circuit 41. Packet switch 1 sends the restored packet CF
is sent back to terminal 2B.

As is clear from the above explanation, according to the present embodiment, the packet switch 1 performs a call setup and a terminal 2
When A returns sendable packet RR (pr),
Thereafter, if the terminal 2B continues to transmit the number of data packets DT (ps) allowed by the window size ws, the buffer management unit 40 confirms that the number q of occupied buffers for the terminal 2A does not exceed the number m of registered buffers. In order to allow packet communication after
(ps) is prevented from transmitting, and as a result, the packet exchange 1 is also prevented from performing invalid processing such as sending the congestion notification packet RNR.

In addition, in this embodiment, the number of occupied buffers q
For example, outgoing packets, incoming packets,
There are many other variations that can be considered, such as managing congestion by retaining each relay packet.
In either case, the effects of the present invention remain the same. Further, FIGS. 3 and 4 are also only one embodiment of the present invention, and for example, the configuration of the buffer management section 40 is not limited to that shown in the figures, and many other modifications may be considered. However, the effects of the present invention remain the same in either case. Furthermore, the configuration of the packet switching network that is the object of the present invention is not limited to that shown in the drawings, and many other variations such as being configured with a plurality of packet switching devices 1 are possible, but in any case, However, the effect of the present invention remains unchanged.

(f) Effects of the Invention As described above, according to the present invention, in the packet switching network, the invalidation processing of the packet switching equipment and terminals due to the congestion of the receiving buffer is reduced, and the efficiency of the packet switching network is improved.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing an example of a conventional reception buffer management method, FIG. 2 is a diagram showing an example of the packet transfer process in FIG. 1, and FIG. 3 is a diagram showing an example of a reception buffer management method according to an embodiment of the present invention. 4 are diagrams showing an example of the packet transfer process in FIG. 3. In the figure, 1 is a packet switch, 2 is a terminal,
3 is Batsuhua, 4 and 40 are Batsuhua management department,
41 is a packet analysis circuit, 42 is a packet sending circuit, 43 is a buffer management circuit, 44 is a timing circuit, 45 is a timing registration circuit, 46 is a selection circuit, CR is a call request packet, CN is an incoming packet, and CA is an incoming call possible. packet, CC is a communicable packet, DT is a data packet, RR is a sendable packet, RNR is a congestion notification packet, CQ is a recovery request packet, CI is a disconnection instruction packet, CF is a disconnection confirmation packet, ps is a transmission sequence number, and pr is a The reception sequence number, ws, is the window size, q is the number of occupied buffers, and m is the number of registered buffers.

Claims (1)

[Claims] 1 Number of buffers m that can be occupied uniquely to the receiving terminal 2A
is registered in advance, and within the range m of the registered buffer number, in the packet switching network 1 that allows the sending of packets from the originating terminal 2B based on the window size ws, when a call is connected to the originating terminal 2B. , the window size ws is added in advance to the number of buffers q occupied by the receiving terminal 2A, and the receiving sequence number pr is determined from the signal RR indicating the packet receivable state sent from the receiving terminal 2A during communication. extract,
The number b of packets whose delivery has been confirmed is calculated from the reception sequence number pr, and the sum (q+b) of the number q of buffers occupied by the receiving terminal 2A and the number b of confirmed packets is the number m of registered buffers.
, the occupied buffer number q is the added value (q
+b) and means for transmitting the transmission ready signal RR.