Using genetic algorithms to improve the reliability of dual homed wireless critical services

Abstract

The wireless access to any service in different contexts is nowadays taken for granted. However, the dependability requirements are different for various services and contexts. Critical services put high requirement on the service reliability, i.e., the probability of no service interruption should be close to one. Dual homing may be used to increase the service reliability in a multi technology, multi operator wireless environment, where the user's mobility necessitates access point selections and handovers. To allow the user to assess the risk of the service session, a prediction of the service reliability is necessary. This prediction must fulfil the need for the optimal sequence of access point selections and handovers with regard to service reliability and being computation efficient to accomplish the need for the real-time operation. We demonstrate how genetic algorithms (GA) may be used to predict and to improve the (near) optimal service reliability by fast and simple heuristics, far more computationally efficient than an Integer Linear Programming (ILP) optimization.