Back to Programs

The Structured Testing Methodology for Software Quality Analyses of Networking Systems
Vladimir V. Riabov, Associate Professor
Rivier College, Department of Mathematics & Computer Science
420 S. Main Street Nashua, New Hampshire 03060, USA
vriabov@rivier.edu
www.rivier.edu/faculty/vriabov/

The metric-based testing strategies have been developed for three networking systems: the Carrier Internetworking solution that manages IP-VPNs over a multiservice ATM infrastructure; Carrier Networks Support system with switches and routers; and Aggregation System for networking services (IP-VPNs, Firewalls, NAT, QoS, and Web steering).

The graph-based metrics (cyclomatic complexity, essential complexity, module design complexity, system design complexity, and system integration complexity) have been applied for studying the decision-structure complexity of code modules, code quality, and the estimated number of integration and unit tests. The C-code (2,447 modules, 149,094 lines) have been analyzed for BGP, Frame Relay, IGMP, IP, ISIS, OSPF, PPP, RIP, and SNMP protocols. 511 modules (19.4%) were founded as unreliable and unmaintainable, including 27% of BGP, IP, and OSPF modules.

Only the Frame-Relay code part is well designed and programmed with few possible errors. The number of unreliable code modules (29%) correlates well with the number of customer requests, error-fixing submits, and the possible errors (1,473) that were estimated with the Halstead’s metrics. Comparing different code releases, it is found that the reduction of the code complexity leads to significant reduction of errors and maintainability efforts. The test/code coverage issues for embedded networking systems are also addressed.

Presenter:

Vladimir V. Riabov, associate professor of Computer Science, teaches Networking Technologies, Software Engineering, and System Simulation and Modeling at Rivier College. His research interests include Complexity Code Studies, Networking Software Test Strategies, Object-Oriented Embedded System Design, and Computational Algorithms. Vladimir received a Ph.D. in Applied Mathematics and Physics from Moscow Institute of Physics and Technology and M.S. in Computer Information Systems from Southern New Hampshire University. He is a member of ACM, AIAA, IEEE, and MAA

Back to Programs