A GA-Based Approach to Resource Scheduling Supporting Flexible Quality Management of Ubiquitous Services

Abstract

In ubiquitous services, concurrent requests from various services for limited service resources such as network bandwidth, easily lead to a problem of resource insufficiency. The resource scheduling for ubiquitous services is the key to improve the tradeoff between request admittance, resource utilization and service quality. In this paper, a GA-based approach to resource scheduling to enable a flexible quality management of ubiquitous services is proposed to solve the problem mentioned above. First, the relationships between service of quality and resource requirements are explored. There are four different types of relations including (1) linear with saturation (LWS), (2) linear with dead zone and saturation (LWDS), (3) shifted step (SS), and (4) exponential (EX). Based on the derivation of the resource-quality model with the four relations, we define the maximum and minimum of resource requirement and regard the scope as the negotiation criterion for quality guarantee in genetic algorithm. Experimental results show that the proposed approach definitely benefits quality guarantee of service and the increasing of service request admittance ratio.