MOBILE AGENTS BASED NETWORK MONITORING

Abstract

Fast growth of computer networks requires efficient network management and monitoring tools for better utilization of resources. Most network devices, nowadays, employ Simple Network Management Protocol (SNMP) agents for network management based on traditional centralized client/server architecture. Such centralized Network Management Systems (NMS) are not adapted to wide spectrum of heterogeneous network installations and configurations. Major problem areas are heterogeneity, complex topologies, scalability and limited bandwidth constraints etc. Present research activity is centered around-providing distributed network management oriented intelligence to different network components. Mobile Agent (MA) paradigm seems to offer a good solution in this area. The main objective of this dissertation is to develop a Network Monitoring Application (IITRNetMonitor) based on Hybrid Models (Mobile Agents integrated with SNMP) providing enhanced functionality with better network latency and least network overhead. Several novel ideas have been implemented to bridge research voids in this area. A new Itinerary Partitioning Approach has been proposed to combat bloated state problem in Mobile Agent Frameworks. We also present a novel approach to write String based network health functions and SNMP table filtering expressions. A new concept of customized reliable traps based on health functions is introduced. IITRNetMonitor provides a mobile network manager with Mobile Agent Generator (MAG) functionality. Generated MA can aggregate SNMP data, semantically compress and intelligently filter it. Our implementation also introduces a new 3-dimensional network state concept for enhancing network monitoring functionality.