TASS: Timing Analyzer of Scenario-Based Specifications | ||||
|
|
GSM ExampleThe next example is the Global System for Mobile Communication (GSM) system [2]. The specification is depicted in Figure 2 which is a loop-unlimited SBS.
Scenario-based specification for the GSM exampleThe GSM system consists of eight components. They are phone users (process User), the mobile phone (process Mobile), two Base Station Systems (process BSS and BSS2), two Mobile Switching Centers (process MSC and MSC2), the Public Switched Telephone Network (process PSTN) and the GSM Common Equipment (process GSM). We assume that each MSC has just one BSS, that is MSC is responsible for BSS, and MSC2 is responsible for BSS2. The user first turns on the mobile, and the mobile checks the signal strength. Then the system goes in three ways. When it is idle, the BSS will check the signal strength in every two seconds (ωt-t1=2, SD Check_Signal), and when the strength is poor, the mobile requests to update the location information. The second flow is the user makes a call the third one is the user receives a page. In both cases a radio resource connection should be established, and the channel would be ciphered. The mobile switches from signaling to voice, after the MSC allocates a voice circuit between BSS and MSC. While on conversation, the mobile reports the signal strength continuously to the MSC via BSS for every ten seconds. And when the signal strength is poor , the call connection would be passed from one MSC to another. If the connection has not been established or the call has not been setup, the mobile returns to its idle state in [2, 3] seconds (2≤ωe-e1≤3, SD Request_Fail; 2≤ωj-j1≤3, SD Refuse_Answer; 2≤ωr-r1≤3, SD Cut_Off). The mobile also returns to idle state in [2, 3] seconds after the connection is released (2≤ωo-oa≤3, SD Connection_Release). The user can turn off the mobile while it is idle. We add many constraints to the system. The user expects the mobile to display necessary indications within one second after it receives a message from the network (0≤h6-h4≤1, SD Call_Setup; 0≤ig-ie≤1, SD Mode_Modify; 0≤q6-q3≤1, SD Page_Setup). The user also expects the mobile to executes an command within one second after the user pushes down the buttons (0≤n3-n1≤1, SD Call_Release; 0≤s3-s1≤1, SD Answer). The specification also describes the following constraints:
We assume each User-Mobile communication takes at least 1 second, each GSM-BSS communication takes at least T1 seconds, each GSM-MSC communication takes at least T2 seconds. And all the other communication time between two processes that we have not mentioned explicitly above is at least 0.5 second.
Figure 2: Scenario-based specification for the GSM example For timing analysis of the GSM specification depicted in Figure 2, TASS is used to solve the following problems:
Experiment DataWe created four cases with various values for W, B1, B2, T1, and T2. All the experiments were conducted on a Pentium 4/2.2GHz/2.0GB PC. We selected some sample results from the experiment. Below are the results from timing analysis with synchronous composition semantics. Table 6 is the results of loop-unlimited timing analysis of GSM SBS. The results show that TASS runs comfortably for checking SBSs with 32 SDs in split seconds.
*:the verification problem disappears because the node is unreachable. Table 6: Sample results of loop-unlimited timing analysis of the GSM specification We also evaluated the bounded timing analysis of GSM SBS with case 4. We set k to 30 so that the number of paths checked was beyond four hundred thousand. Moreover we observed that during the computation TASS consumed less than 50MB of memory. This advantage comes from our DFS manner in traversing the specifications. Since TASS can do the analysis without manual interference, given enough time it is possible to check a large amount of paths.
Table 7. Sample results of bounded timing analysis of the GSM specification
[2] "Telecom Message Sequence Charts", EventHelix.com Inc. http://www.eventhelix.com/EventStudio/telecom_message_sequence_charts/ |
©2011 SEG CS NJU | Introduction Download Case Study 1 Case Study 2 |