Research On Soft Load Balancing Algorithms Over Heterogeneous Network

In recent years, field of wireless communications has achieved unprecedenteddevelopment, people continue to put forward a variety of innovative wirelesscommunication technology, for a time, a variety of formats for heterogeneous wirelessnetworks. Meanwhile, with the growing popularity of mobile Internet and the rapiddevelopment of smart terminals, especially the rapid development of mobile endmultimedia applications, traffic moving side showing explosive growth. The single-modecommunication network due to the lower-end throughput, latency is large, has beenunable to meet existing demand. It will be a large number of heterogeneous wirelessnetworks overlapping coverage scenarios in the future, in this kind of scenario, the use ofthe complementary advantages of heterogeneous wireless networks, is one of theeffective means to solve these problems. Therefore, it is necessary to study acollaborative sharing technology under heterogeneous wireless networks from theperspective of resource utilization to enhance the overall integration of heterogeneouswireless network system, the load balancing is clearly a very effective means. The focusof our study was to soft load balancing strategy.Under this policy,the data traffic of a useris divided into subflows, each of which flows into a different access network in order toachieve load balancing in heterogeneous wireless network integration system.First，we propose an optimal allocation mechanism for heterogeneous wirelessnetwork traffic in the network environment. A plurality of heterogeneous networks canbe seen as a combination of end-user services can be simultaneously with the networkconfiguration. For simplicity, each of the network are modeled as a M/M/1queue. Ourgoal is to minimize the number of jobs in each queue bound. The optimal solution of theproblem can be solved by Lagrange multiplier. According to various network service rate,the optimal data rate of the generated assigned to each network. Simulation results show that the algorithm than the average allocation algorithm and greedy algorithm has higherthroughput, improved spectral efficiency.When a parallel link transmission of multimedia services and applications in realtime business, because these two types of business aresensitive to data packet delay, jitterand packet reordering, therefore, in order to improve the utilization of parallel links,studying a kind of effective load distribution model based on the delay controlled iscrucial. We propose a load distribution strategy based on link delay controlled calledDCSLB algorithms. The purpose of DCSLB algorithm in heterogeneous wirelessnetwork integration scene, consider the bandwidth and propagation delay factor of themulti-link, and adjust the network load ratio(the ratio of split) assigned to each link bydynamically to minimized the maximum link delay. On the one hand, reducing thereceiving end packet reordering wait time; on the other hand, makes end to end delay ofeach link is minimum, thereby reducing the probability of receiving end packetreordering, thus making the delay of the system successfully transmiting of a packet isreduced.Subsequently, we built a verification system based on the integration of LTE andWLAN to implement DCSLB algorithm.Finally, we have expanded the soft load balancing scenario. Aim at future macro celland small cell cellular network coverage and collaborative WLAN AP mesh networkcomposed scenes, we have presented a hierarchical semi-centralized network architecture,it can effectively reduce the complexity of managing radio resources management server,through the introduction of spare radio resource management server so that theheterogeneous wireless network have disaster recovery capabilities.