Research On Edge Caching Via Game Theory In Fog Radio Access Network

Driven by the dramatic growth of intelligent devices and mobile applications,wireless networks have been suffering an unprecedented data traffic pressure in recent years.Fog radio access networks(F-RANs)can effectively accommodate the data traffic pressure and improve the quality of service by edge caching.However,the conflict between increasing content demands and limited communication resources restricts the development of edge caching.Applying game theory to edge caching is an effective way to solve this problem.In view of these problems,the thesis studies the edge caching via game theory in F-RANs.Firstly,the non-uniform pricing based edge caching content allocation problem is studied.We formulate a Stackelberg game where the cloud server owning contents acts as a leader and the fog access point(FAP)clusters renting contents are the followers.Then,the profit functions of the cloud server and the F-AP clusters are formulated,respectively.By using the backward induction method,the optimal strategies are obtained to achieve the Nash equilibrium(NE).Besides,we also investigate the condition to achieve the NE.Numerical results are provided for verifying the efficiency of the proposed resource allocation strategy and the simulations also show the effects of the F-AP's storage capacity on the resource allocation strategy.Secondly,the non-uniform pricing based edge caching resource allocation problem is investigated.To motivate content providers(CPs)to participate in this resource allocation procedure,we introduce an incentive mechanism.The cloud server sets non-uniform prices of F-APs and leases them to the CPs,while the CPs cache contents in leasing F-APs and get rewarded by the raised content hit rate.We formulate the interaction between the cloud server and the CPs as a Stackelberg game and solve the optimization problems to achieve NE.By exploiting the multiplier penalty function method,the constrained optimization problem for the cloud server is transformed into an equivalent non-constrained optimization problem.Then,we propose the pricing algorithm to solve the non-constrained optimization problem by applying the simplex search method.The existence,uniqueness and the Pareto optimality of the NE are proven.Simulation results show the rapid convergence of the proposed algorithm and the superiority performance in improving content hit rate.Finally,the uniform pricing based and global optimization based edge caching resource allocation problems are investigated.First,the uniform pricing based resource allocation is studied.In this circumstance,all the CPs participating in the game are considered as a group,and the cloud server charges F-APs with same price to them.The competition among the CPs is then eliminated.Meanwhile,the relationship between the pricing strategy of the cloud server and the nature of the CPs is also studied.Second,a global optimization based resource allocation strategy is achieved by using the proposed sub-gradient based resource allocation algorithm.Simulation results show that the uniform based resource allocation strategy has lower computational complexity,and the proposed sub-gradient based resource allocation algorithm can converge quickly,and the obtained resource allocation strategy can achieve the highest average content hit rate.