Generic Log Analyzer

Abstract: In the field of mobile telecommunication to meet demanding end user requirements in terms of quality of experience and service, different protocol specifications have been defined ranging from GSM to WCDMA and now LTE which includes LTE-Advanced. Specifications of these protocols are being revised on a frequent basis and thus bringing more challenges at development side. For the development teams, tracking and analyzing the cause of failure in protocol stacks is a daunting task and consumes valuable engineering resources where they have to use conventional debugging methods. One widely used conventional method till now is to generate logs and ask an expert to analyze and interpret issues against protocol specifications by scanning through hundreds of lines of traces.
This master thesis describes an implementation of a Generic Log Analyzer which can ease the analysis of logs generated across the fields. The experiment to verify the validity of the log analyzer was done using two different log formats. The first one was a formal and consistent format and most of the information was coded. It was learned that the log analyzer can point out missing signals/messages from the scenarios in log files after decoding the log messages. Two scenarios were selected for formal file format, namely Mobile Originated Call and Mobile Terminated Call from 3GPP specifications.
The second selected file format was informal and inconsistent because there was no standard defined format for log entries and one log file can have multiple types of message formats. It was learned that the log analyzer not only filtered out required messages but also reports problematic areas in the selected scenario where the mobile phone transitions from one to another state frequently. The scenario selected for the informal type of format was “Power on” of a mobile device as per defined in the 3GPP specifications.
After the experiments were conducted, it was learned that such a tool can be extended to domain specific rules to add more intelligence if the rule framework was provided for the end user. Thus, such a tool can not only consolidate experience at a single place and help intended users to find minor to advanced level problems in the log files. It can also act as a watchdog to ensure system quality by reporting level of test coverage based on finding scenarios from logs generated during the system test.