Admission Control Algorithms In Large-Scale Wireless Communication Networks

With the rapid development of modern wireless communication networks,the number of users is increasing sharply and inevitably,so there is an emerging risk of infeasibility for network management problem,which is urgent today.That is,the Quality of Service(Qo S)requirement cannot be satisfied for all the users due to the interference induced from multiple concurrent transmissions and the limited transmit power budget.In consequence,the whole network breaks down and no user can be served,even if the infeasibility is caused by just few users.Obviously,a more reasonable strategy is that instead of simply claiming infeasibility,the network turns to serving as many users as possible with the predefined Qo S constraints.User admission control is a potential solution to these problems.Actually,many works have addressed the application of user admission control in network management.Since we have read quite a few relative references comprehensively,we finally found two valuable research subjects,which have not been reported yet.Fisrt,it's worth to mention that most of admission control problems are power minimization problems.In large-scale communication networks that deploy a large number of BSs,however,the maintenance power that keeps device in good conditions cannot be ignored anymore.Therefore,the total power consumption has to cover the maintenance power.BS activation,i.e.,appropriately associating users to just a few BSs and deact ivating the rest,is proposed as a promising solution to limiting the maintenance power.Thus,joint BS activation and beamforming is performed to balance the maintenance power and the transmit power effectively.However,although user admission control and BS activation have been studied intensively,they are seldom jointly considered,so we integrate user admission control into the power minimization problem based on joint BS activation and beamforming to minimize the total power consumption.Compared with the strategy without user admission control,the proposed one has two attractive advantages.First,integrating user admission control helps identify the infeasible users,and then the remaining users can still be served after the infeasible users are rejected.Second,admission control also selects the users whose Qo S requirements can be easily satisfied with a relatively low power cost,thus improving the power efficiency.Second,we also note that the known optimization problems based on user admission control and even our previous research problem are basically power minimization or user quantity maximization problems under given Qo S constraints.Besides,there are also user sum-rate maximization and max-min fairness problems in network management.Actually,the problems based on sum-rate maximization itself can select users to join the network,but it can't guarantee the fairness among the selected users.Yet the max-min fairness networks can guarantee it,however,the networks alone cannot implement the user selection,so it is reasonable to consider user admission control in max-min fairness networks.In max-min fairness networks,if all users are allowed to have an access to the network,it will definitely go to the situation where only a very low max-min SINR could be achieved,due to an enormous number of users whose achievable SINR is totally different,which is considered meaningless.Therefore,its low power efficiency results in an obvious conflict with the principles of green communication.Based on these considerations,it motivates us to explore the user admission control in the max-min fairness networks.Morover,due to the challenging programming with NP-hardness,we seek for some efficient approximation to both of the original joint optimization problems.To this end,we first recast the original problems as a convex sparse optimization problem and an SOCP problem respectively,and then design a low-complexity algorithm to solve them efficiently.Specifically,in large-scale communication networks,we consider solving the problems in a distributive way.Motivated by this,we finally develop a distributed algorithm to solve the problem iteratively,utilizing the alternating direction method of multipliers(ADMM).Interestingly,closed-form solutions could be derived in each iteration,thus giving the algorithm a very low complexity.