Network and Service Failure Restoration and Prevention in Multi-hop Wireless and Mobile Networks

Abstract

Wireless networks are more prone to failures than their wireline counterparts. The unique characteristics of wireless networks introduce fundamental challenges to the design of restoration related wireless networks that can satisfy the performance and reliability requirements. Network and service restoration schemes posed for wireline networks (such as the Internet) are poorly suited for highly dynamic and unstable wireless networks. This dissertation investigates the network and service restoration as well as QoS support issues in the design of reliable multi-hop wireless networks.

It first investigates the network restoration problem that achieves minimum performance degradation in multi-hop wireless stationary networks under node failures and jamming attacks. The proposed defense strategy dynamically adjusts the channel assignment and traffic routes to bypass the failed nodes and the jamming area. We apply an optimization-based approach that formulates network restoration strategies as linear programming problems. Based on the LP solutions, we provide a greedy scheduling algorithm using both dynamic and static channel assignments. Network performance of these optimal network restoration strategies is evaluated via comprehensive simulation study under different failure scenarios.

It second investigates the service composition and restoration problem that achieves minimum service disruptions in multi-hop wireless mobile networks under frequent mobility-caused wireless link failures. To address this issue, we propose a service composition and restoration framework and formulate the problem of minimum-disruption service composition and restoration (MDSCR) as a dynamic programming problem. We then present our MDSCR heuristic algorithm that approximates the optimal solution with one-step lookahead prediction. Comprehensive simulation study is conducted to analyze the performance of our MDSCR algorithm under the impacts of service path length, service link length requirement, traffic type, service component redundancy, system dynamics, and disruption penalty function.

It also investigates the QoS support scheme for health monitoring services that integrates XML-based service description, patient admission policy, differentiated scheduling and queue management. The proposed solution is implemented in CareNet, our two-tier wireless sensor system for remote healthcare. Extensive experimental results show that our system can provide low latency and low loss rate assurance to critical medical service traffic.