Mobile data offloading based on minority game theoretic framework

Mobile data offloading is a current-day networking paradigm to channelize certain fraction of the cellular data traffic over unlicensed spectrum of WiFi. In this paper, we present a novel data offloading scheme built upon the exponential learning-based minority game (MG) theory. Performance comparison between cellular and WiFi services with respect to the offered load, has been used to derive an appropriate offloading condition for our proposed MG-based distributed data offloading algorithm. Effectiveness of the MG algorithm is tested by studying its performance through extensive simulation by varying several important parameters, like, pricing parameter \(\beta\), cellular offered throughput (\(S_{{c}}\)), and temperature coefficient of the algorithm (\(\gamma\)). An effective model for tuning pricing parameter \(\beta\) with respect to the offered load (named as, Target Pricing scheme) is also presented using reverse engineering approach by considering the dynamic traffic condition. We have provisioned the application of our algorithm in multi-access point environment. We have studied the behaviour of different classes of nodes in heterogeneous population, while applying a MG-based networking algorithm. Through extensive NS3 based simulation we have evaluated the performance of our proposed algorithm in an IEEE 802.11ax environment and studied the effect of MIMO, QoS, transport layer and unsaturated traffic condition.