Study On Resource Management Heterogeneous Wireless Networks

With prolific growth of wireless data traffic and rapid development of the multipleradio technologies, heterogeneous wireless networks (HWNs) will be the trend infuture. Resource management plays an important role for the integrated variousnetworks. In resource management of heterogeneous wireless networks, the firstproblem is to choose access control schemes, such as the single radio access andmultiple radio access. The different access control schemes will affect the performanceof integration of heterogeneous wireless networks. Accordingly, the traffic allocationwill be disparities with the different access control schemes in heterogeneous wirelessnetworks. It is important to investigate these above problems for resource managementin heterogeneous wireless networks. This dissertation mainly focuses on the accesscontrol scheme and traffic allocation for resource management in heterogeneouswireless networks. We firstly address the effect of the access control schemes onintegrated performance of heterogeneous wireless networks. Then, the traffic allocationalgorithm for multiple radio access is to be design. Finally, we investigate the loadbalancing for interference coordination in heterogeneous networks (HetNets).The dissertation is supported by the National Nature Science Foundation of China(No.61231008), National Basic Research Program of China (No.2009CB320404),National Science Fund (No.61102057), Program for Changjiang Scholars andInnovative Research Team in University (No. IRT0852) and the111Project(No.B08038). This dissertation studies the access control scheme and traffic allocationfor resource allocation and the related interference management in the HWNs scene.The main achievements and results of this dissertation are listed as follows:1. The problem to choose access control schemes in heterogeneous wirelessnetworks is studied. We analyze the performance of three access control policies:optimal radio access technology (O-RAT) selection, the aggregated multi-radio access(A-MRA) technology and the parallel multi-radio access (P-MRA) transmission, wherethe O-RAT means that the incoming traffic always tries to access one network with themaximum service rate before admission. The A-MRA intends to allocate the same datato all networks. And the traffic will leave the system if it is accomplished first by one of these networks. The P-MRA is that the incoming traffic is split into differentnetworks. The traffic is served with the sum of the service rates provided by overallnetworks. First, we consider the different multiple access protocols and service rate ofdifferent networks. Next, the complete sharing discipline is used to model the WLANand the equal sharing discipline is used to model the celluar network, so the servicerate of the two networks can be obtainted. Then, the three access control policies areembedded into the Markov chain, the blocking probability and average delay of threeaccess control policies for HWNs can be obtainted. Simulate results show that theP-MRA policy can obtain the lower delay than the other two policies with the specificparameters of the different multiple access protocols and service rates.2. The traffic allocation for multi-radio access with probabilistic manner is studiedin heterogeneous wireless networks. First, via treating the integrated heterogeneouswireless networks as different coupled and parallel queuing systems from a queuepoint, the average delay in heterogeneous wireless networks is derived in aprobabilistic manner with the delay of single network. Then，the minimal averagedelay is formulated as an objective function with the constraint of queue stable. For thepurpose of illustrating the effectively of our proposed model, the Cellular/WLANinter-working is exploited, a heuristic search algorithm is used to get the optimalprobability of splitting traffic flow to minimize the system average delay. Further, aMarkov process is applied to evaluate the performance of the proposed schemecomparing with the network selection algorithm. Numerical results illustrate that theprobabilistic split algorithm can obtain the lower blocking probability and decrease theaverage delay compared with network selection algorithm.3. The traffic allocation to minimize delay for end to end of parallel multi-radiotransmission is studied in heterogeneous wireless networks. First, we take into accountthe effect of the available bit rate and network delay of different networks on the trafficspiltting, and analyze the transmission delay of the traffic spiltting in differentnetworks. Then, the minimal delay for traffic allocation is formulated as anoptimization problem. Based on Beckmann transformation method, we obtain thethreshold for the optimal traffic allocation. According to such threshold, a minimalsubset of admitted networks can be chosen and traffic allocation among these networkscan be done to achieve the minimum delay. Finally, a joint network selection andtraffic allocation algorithm is proposed for parallel multi-radio transmission.Simulation results show that the proposed algorithm could decrease the average delay of system and guarantee the maximum throughput under heavy load conditioncompared with the existing algorithm effectively.4. In heterogeneous network composed of spectrum-sharing macrocell andpicocell, the joint uplink and downlink of load balancing for eICIC is studied. First, wepropose configuring uplink transmission for the user in macro for almost blanksubframe (ABS) of eICIC to improve the utilization of ABS, refer to UM-ABS. Andwe assume that the user can associate with the different BSs in the downlink anduplink and consider the asymmetry channel model in the two directions. Thus, with theUM-ABS and asymmetry downlink and uplink, we adopt an optimization frameworkto investigate the problem of asymmetry association for downlink and uplink and thetime resource allocation between macrocell and picocell. The optimization problem ismixed integer programming, and then a relaxed-rounding approximate solution isprovided. Numerical results show the proposed algorithm can improve systemthroughput and obtain the rate gain for users relative to the existing approachessignificantly.