End user-oriented node resource allocation: An application-based method

For a network, how to allocate limited resources to achieve high network reliability has been concerned by engineers. From the perspective of the users, the quality of applications is the most intuitive perception of network reliability. Previous allocation methods focus little on applications, which makes it difficult to optimize resource with given application requirements. This paper proposes an application-based resource allocation method during network design. Under given application scenario and resource constraints, the allocation model is constructed to maximize application reliability, which is presented as a novel user-oriented metric to assess the quality of applications. Furthermore, improved genetic algorithm (GA) is adopted to obtain optimal allocation strategy. To improve the efficiency of the algorithm, application-affected node importance is evaluated to generate initial population. Simulation results demonstrate that the proposed method can achieve higher application reliability than the other three allocation methods when applied to an Avionics Full Duplex Switched Ethernet (AFDX) case. This makes it possible to optimize resource allocation in the light of different application demands in network design.