Abstract:

Resource allocation is one of the core techniques in multi-homing delay/disrupt tolerant networking（DTN）, which decides the efficiency of DTN routing, and even affects the whole system. However, schemes based on neural network and genetic algorithms are of computational complexity that is not applied to real-time applications. Therefore the information matrix of access routers and quasi-two-dimensional access bandwidth constraints were designed based on predictable environment of train mobile network, and then an approximation algorithm with O（n） algorithm complexity, based on utility differential, was proposed to meet the real-time applications. Finally simulation showed that call-drop-rate in handover and new call-block-rate only increase by about 1/4, compared with that of the optimized-algorithm, but calculation time decrease down to its 36.99%. It means the algorithm ensures real-time response and could provide good QoS at the same time. The algorithm could be further applied to the real 2-dimensional moving scene.

Key words: resource allocation, mobile internet access, real-time, delay/disrupt tolerant networking, utility function, quality of service