CN100401835C - Method for Eliminating Simultaneous Calls in Distributed Call Processing System - Google Patents

Abstract

The invention discloses a method for canceling simultaneous calls in a distributed call processing system, and defines two states: S0 indicates that a call request has not been received, the call includes a local call and a remote call, and S1 indicates that the call has been made by the local end Call allocation resources, in the S0 state, after the local end successfully allocates resources for the local call and applies for resources from the remote end, it will move to S1, otherwise it will remain in S0; in the S1 state, when the remote end receives a response to the local end call resource request , if the resource is obtained, the local call is accepted, otherwise the local call is released, and all of them are moved to S0. When the resource request of the remote call is received and the resource can be allocated, it is kept as S1, otherwise the remote call has a high priority and is released. When resources can be allocated for the local call, release the local call, allocate resources for the remote call and return a response, and move to S0; otherwise, reject the remote call and remain in the S1 state. The invention eliminates call co-emption, does not reduce the efficiency of call processing, and realizes priority processing of high-priority calls.

Description

Method for Eliminating Simultaneous Calls in Distributed Call Processing System

technical field

The invention relates to a call processing method in the communication field, in particular to a method for eliminating call co-emption in a distributed call processing system.

Background technique

In the existing signaling subsystem, centralized processing is a commonly used call processing method. The advantage of centralized call processing is that all the resources required for call establishment are allocated on one processor, the control is simple, and it can avoid the phenomenon of call grabbing that may occur when resources are allocated on multiple processors. However, due to the centralized Call processing is a linear operation and cannot be processed concurrently, so its call processing capability is relatively low. Distributed call processing can operate concurrently, and multiple processors in the system process call requests from different ports at the same time, so its call processing capability is much higher than that of centralized processing, but distributed call processing also inevitably brings The problem of calling and grabbing is solved.

The occurrence of call grabbing is based on the following three prerequisites: A. Distributed system; B. Two call requests reach their respective processing modules at the same time, and at the same time require the resources of the peer module; C. The remaining resources on the two processing modules At least one of the resources is less than the sum of the resources requested by the two calls, but either call can be accommodated.

Taking the system shown in Figure 1 as an example, RR1 and RR2 represent the remaining resources on modules _M1 and _M2 respectively, and _C1x and _C2y represent call requests from ports _P1x and _P2y of M1 and _{M2 respectively} . If the resources requested by C _1x and C _2y are R _1x and R _1y respectively, the premise C can be expressed as formula (1):

max(R1x, R1y)≤min(RR ₁ , RR ₂ )＜R _1x +R _1y (1)

If C _1x and C _2y arrive at M ₁ and M ₂ at the same time, although a call from C _1x and C _2y should be accepted at this time, in fact, since each call is allocated local resources on the ingress module first, As a result, all calls from the remote end are rejected by the local end. If these two call requests are constantly required to be set up, then this phenomenon will continue continuously, and if the priorities of the two call requests are inconsistent, the high-priority call will be interfered by the low-priority call.

At present, some systems adopt centralized management of resources to eliminate call co-emption, but this will reduce call processing efficiency. In the ATM (Asynchronous Transfer Mode) network, the signaling system standard follows ITU-T's Q.2931 series of protocols or the ATM Forum's UNI/NNI (User Network Interface/Network Network Interface) series of protocols, but these protocols have not proposed A solution to the simultaneous snatching of calls.

Contents of the invention

The technical problem to be solved by the present invention is to provide a method for eliminating simultaneous calls in a distributed call processing system, which does not reduce the efficiency of call processing and simultaneously solves the problem of simultaneous calls.

In order to solve the above-mentioned technical problems, the present invention provides a method for eliminating simultaneous calls in a distributed call processing system, which is characterized in that:

In each processing module, define two states of the state machine for calling together: S0: indicates that the call request has not been received, and the call includes the local call and the remote call; and S1: indicates that resources have been allocated for the local call, and will The calling and grabbing state machine is embedded in the standard state machine to form a dual state machine model;

In the S0 state, the processing module performs operations in the following manner: after the local call reaches the local module, the local module allocates resources for it, and if the resources are allocated successfully, it applies for resources from the remote module for the local call, and at the same time The call state is transferred from S0 to S1; if resource allocation is unsuccessful, the local call is rejected and the call state remains at S0;

In the S1 state, the processing module performs operations in the following manner:

If the local module receives the remote module’s response to the local call resource request, it judges whether the resource is obtained in the remote module, if yes, accepts the local call, and transfers the call state from S1 to S0 at the same time, if the resource is not obtained, then Release the call at the local end, and transfer the call state from S1 to S0 at the same time;

If the local module receives the resource request from the remote call to the local module, if the resource request can be satisfied, it will allocate resources for it and return a response, and keep the call status as S1; if it cannot be satisfied, the priority of the remote call is higher than When the local call is released and the remote call resource request can be satisfied after the local call resource is released, the local call is released, resources are allocated for the remote call and a response is returned, and the call state is moved to S0 at the same time. When the resource request of the remote call cannot be satisfied, the remote call is rejected and a response is returned, and the call state is kept as S1.

In the ATM network, the level of the priority can be confirmed by the following method: first sorted by the quality of service level, from high to low: constant bit rate, real-time variable bit rate, non-real-time variable bit rate, available bit rate rate, unspecified bit rate; when the QoS levels are the same, calls with small resource requirements are given priority; if both the QoS level and resource requirements are the same, calls initiated by ports with small global port indexes are given priority.

The processing module can use a call priority dynamic queue to store multiple calls in progress, and when call simultaneous grabbing occurs, it will process the calls in the queue sequentially according to the order of priority from low to high according to the operation method in the S1 state. call. In this way, when a call in the queue has a resource problem, the low-priority call can be rejected first, improving the processing efficiency.

In the above method, in the S1 state, when the remote call resource request can be satisfied after the local call resource is released and the priority of the remote call is lower than that of the local call, the remote call can be rejected and a response can be returned, and the call status can be kept as S1; or after receiving the response from the remote end to the call resource request of the local end, if the resource is obtained, reject the remote call and return a response, accept the call at the local end, and transfer the call state to S0 at the same time; if the resource is not obtained, release the local end Call, allocate resources for the remote call and return a response, and move the state machine to S0. The latter method can accept the remote call when the priority of the remote call is low and the local end can meet its resource request, but the remote end cannot meet the resource request of the local call.

It can be seen from the above that the present invention effectively solves the problem of call co-emption in the signaling system by implementing a finite state machine based on call priority order processing, and still adopts a distributed resource management method, without reducing the efficiency of call processing . At the same time, it realizes the priority processing of high-priority calls, makes the call processing have the quality of service (QoS) feature, and makes full use of limited network resources.

Description of drawings

FIG. 1 is a schematic diagram of call co-grabbing in a system composed of two processing modules.

Fig. 2 is the situation that call grabbing takes place on the ATM node of the embodiment of the present invention.

Fig. 3 is the call occurrence process embedded in the call co-emption state machine on the ATM node of the embodiment of the present invention.

Fig. 4 is the call acceptance process embedded in the call co-emption state machine on the ATM node of the embodiment of the present invention.

Fig. 5 is a schematic diagram of a call priority dynamic queue according to an embodiment of the present invention.

Detailed ways

The invention implements the finite state machine based on the order processing of the call priority to realize the priority processing of the high-priority call, and eliminates call co-emption at the same time. Firstly, two states of the state machine for calling and snatching are defined: S0 and S1; their meanings are as follows:

S0-NULL state: Indicates that the module has not received a call request;

S1-ALLOCATED state: Indicates that the module has allocated resources for calls arriving at this module.

Taking the network-side processing process of the broadband network signaling system as an example, the specific implementation manner of the present invention will be further described in conjunction with the accompanying drawings.

On the ATM node shown in Figure 2, both sides of the ATM switch are connected with the line interface card 1 and the line interface card 2 respectively, and the SETUP messages of calling 1 and calling 2 arrive at the line interface card 1 and the line interface card 2 respectively.

According to the call processing flow of Q.2931, when a call occurs, if the local resources can be allocated for the call, the call state is transferred from the zero state (N0) to the call state (N3). In this standard state machine, the calling and grabbing state machine of the present invention is embedded to form a dual state machine model. Before receiving the SETUP message, the call state is N0+S0, and when the standard call state changes from N0 to N3, the S0 state is moved to S1 at the same time.

Line interface card 1 calls generation processing flow (being the processing flow under S0 state) to call 1 as shown in Figure 3, comprises the following steps,

Step 110, after receiving the SETUP message, decode it first, if successful, execute the next step, otherwise execute step 150;

Step 120, perform call admission control (CAC), if successful, perform the next step, otherwise perform step 150;

Step 130, distribute virtual path identifier/virtual channel identifier (VPI/VCI), if successful, perform the next step, otherwise perform step 150;

Step 140, apply for resources from the line interface card 2, move the state machine to N3+S1, and end.

Step 150, release the call, keep the state machine as N0+S0, and end.

The CAC algorithm of the present embodiment uses the equivalent resource algorithm, and to the calls of different QoS grades, the occupied resources are different in size, and more equivalent resources are allocated to the calls with high QoS grades; while the VPI and VCI allocation methods adopt static The array is implemented by simulating a static linked list, by setting the 0/1 flag in the linked list to indicate whether the VPI/VCI resource is available;

At the same time, the line interface card 2 performs the same operation on the call 2, except that in step 140 it applies for resources from the line interface card 1 .

When the remaining resources on line interface cards 1 and 2, the resources requested by calls 1 and 2 satisfy the relationship of formula (1), that is, at least one of the remaining resources on the two line interface cards is less than the sum of the resources requested by the two calls , but when any call can be accepted, the same grab may occur.

In the call acceptance process of the line interface card, the call same-grab state machine is also embedded in the standard state machine to form a dual-state machine model. In the outgoing call state, the current state machine is N1+S1.

Still take the call admission process (i.e. the processing flow under the S1 state) of the line interface card 1 below as an example to illustrate the method for eliminating the simultaneous snatching of calls in this embodiment, as shown in Figure 4, including the following steps:

Step 210, the local module (referring to the line interface card 1) first receives the resource request response of the remote module (referring to the line interface card 2) to the local call (referring to the call 1);

Step 220, judging whether to obtain resources in the remote module, if yes, execute step 230, otherwise execute step 240;

Step 230, send out the connection message of the local call, move the state machine to N10+S0, and end. N10 indicates that the local call has been accepted, and the local call has entered a stable state;

Step 240, release the call at the local end, move the state machine to N0+S0, and end;

Step 310, the local module first receives the resource request of the remote call (referring to call 2) to the local module;

Step 320, judging whether the remaining resources of the local end meet the requirements of the remote call, if yes, consider that the same grab has not occurred, and execute step 380, otherwise, execute the next step;

Step 330, judging whether the priority of the remote call is higher than that of the local call, if yes, execute the next step, otherwise, execute step 370;

Step 340, after judging the release of the local call, whether the remaining resources of the local end can satisfy the remote call, if yes, perform the next step, otherwise, perform step 370;

Step 350, releasing the call resources of the local end and clearing the call of the local end;

Step 360, allocate resources for the remote call and return a response message, move the state machine to N0+S0, and end;

Step 370, rejecting the remote call and returning a response message, keeping the state machine as N3+S1, and ending;

Step 380, allocate resources for the remote call, keep the state machine as N3+S1, and end.

In step 330, a priority algorithm as defined below may be used:

First sort by the quality of service (QoS) level, such as CBR (constant bit rate) > rtVBR (real-time variable bit rate) > nrtVBR (non-real-time variable bit rate) > ABR (available bit rate) > UBR (unspecified bit rate) );

When the QoS levels are the same, the calls with small resource requirements are given priority;

If the QoS levels and resource requirements of the two calls are the same, they are sorted according to the global index of the port, and the call initiated by the port with the lower global index of the port is given priority.

If it is on an IP network, the TOS field of the IP packet can be used to determine the priority.

The call acceptance process on the line interface card 2 is the same as that of the line interface card 1, except that the meanings of the local and remote modules and the local and remote calls are different. Combining the processing of the two line cards, it can be seen that, when the simultaneous grabbing of calls occurs, assuming that the priority of call 1 is higher than that of call 2, then line interface card 1 will run to step 370, rejecting the resource request of call 2, and at the same time, Line interface card 2 will run to step 360, release the resources of call 2, allocate resources for call 1 and return a response; next, after line interface card 1 receives the response that line interface card 2 allocates resources for call 1, it will A connect message for call 1 is sent, so that call 1 is finally accepted and call 2 is released. Conversely, if the priority of call 1 is lower than that of call 2, call 2 will be accepted and call 1 will be released, so that when call co-preemption occurs, call co-preemption is eliminated and high-priority calls are prioritized.

There may be multiple calls in progress in one module at the same time, that is, calls that have not reached the steady state N10. The embodiment of the present invention establishes a call priority dynamic queue to save the calls in progress. As shown in Figure 5, when the call When it occurs, according to the QoS level of the call, it will be sorted in the order of UBR, ABR, nrtVBR, rtVBR, and CBR from low to high, and the call will be enqueued. When call co-preemption occurs, the calls are processed sequentially according to the above-mentioned call state machine processing process in order of priority from low to high. In this way, when a call in the queue has a resource problem, the low priority call can be rejected first, which can Improve processing efficiency. When the call is complete, dequeue the call from the queue.

On the basis of the above-mentioned embodiments, some transformations can also be made. For example, after the resources of the local end cannot satisfy the remote call in step 320 of the embodiment, the following steps are performed:

After judging the release of the local call, whether the remaining resources of the local end can satisfy the remote call, if not, reject the remote call and return a response message, keep the state machine as N3+S1, and end; if yes, execute the next step;

Determine whether the priority of the remote call is higher than that of the local call. If so, release the local call, allocate resources for the remote call and return a response message, and move the state machine to N0+S0; if the priority of the remote call is low, Execute the next step;

After receiving the resource request response from the remote module for the local call, if the resource is obtained, then reject the remote call and return a response message, send a connection message for the local call, and move the state to N10+S0 at the same time; if the resource is not obtained , release the local call, allocate resources for the remote call and return a response message, and move the state machine to N0+S0.

The difference between the above steps and the embodiment is that when the remote call has a low priority and the local end can satisfy its resource request, and the remote end cannot satisfy the resource request of the local call, the remote call can be accepted. However, according to the flow of the embodiment, both the remote call and the local call are rejected. However, both methods can eliminate call co-preemption when the call co-preemption meeting the condition of formula (1) occurs.

Claims (4)

1. a method for eliminating calls in a distributed call processing system, characterized in that: In each processing module, define two states of the state machine for calling together: S0: indicates that the call request has not been received, and the call includes the local call and the remote call; and S1: indicates that resources have been allocated for the local call, and will The calling and grabbing state machine is embedded in the standard state machine to form a dual state machine model; In the S0 state, the processing module performs operations in the following manner: after the local call reaches the local module, the local module allocates resources for it, and if the resources are allocated successfully, it applies for resources from the remote module for the local call, and at the same time The call state is transferred from S0 to S1; if resource allocation is unsuccessful, the local call is rejected and the call state remains at S0; In the S1 state, the processing module performs operations in the following manner: If the local module receives the remote module’s response to the local call resource request, it judges whether the resource is obtained in the remote module, if yes, accepts the local call, and transfers the call state from S1 to S0 at the same time, if the resource is not obtained, then Release the call at the local end, and transfer the call state from S1 to S0 at the same time; If the local module receives the resource request from the remote call to the local module, if the resource request can be satisfied, it will allocate resources for it and return a response, and keep the call status as S1; if it cannot be satisfied, the priority of the remote call is higher than When the local call is released and the remote call resource request can be satisfied after the local call resource is released, the local call is released, resources are allocated for the remote call and a response is returned, and the call state is moved to S0 at the same time. When the resource request of the remote call cannot be satisfied, the remote call is rejected and a response is returned, and the call state is kept as S1. 2. call as claimed in claim 1 and eliminate method simultaneously, it is characterized in that, in ATM network, the height of described priority is confirmed by following method: First sort by the quality of service level, from high to low: constant bit rate, real-time variable bit rate, non-real-time variable bit rate, available bit rate, unspecified bit rate; when the quality of service level is the same, the resources are given priority Calls with small requirements; if the quality of service level and resource requirements are the same, the calls initiated by the port with the small global index of the port are given priority. 3. The call elimination method as claimed in claim 1 or 2, characterized in that, the processing module adopts a call priority dynamic queue to save a plurality of calls in progress, and when a call occurs, the call will be processed according to the priority. From low to high, the calls in the queue are processed sequentially according to the operation method in the S1 state. 4. The method for eliminating call co-emption as claimed in claim 1, characterized in that, in the S1 state, the remote call resource request can be satisfied after the local call resource is released and the priority of the remote call is lower than that of the local call , reject the remote call and return a response, and keep the call status as S1; or after receiving the remote response to the local call resource request, if resources are obtained, reject the remote call and return a response, accept the local call, and at the same time set The call state moves to S0; if the resource is not obtained, release the local call, allocate resources for the remote call and return a response, and move the state machine to S0 at the same time.

Images