جدول هش توزیع شده سلسله مراتبی و تکرار متناسب مبتنی بر مدیریت تحرک در مقیاس بزرگ اینترنت تلفن همراه

A considerable amount of research on distributed hash table (DHT) based mobility support schemes, which are highly user scalable and load balanced, has been done. But these schemes have shortcomings in query performances and network scalability. It is because although routing of overlay is effective, there is inconsistency between logical and physical topologies, so the actual physical network performances are not necessarily efficient. In this paper, we introduce a replication technology based DHT-based mobility support. Then all queries from any subnet can get responses as early as possible, i.e., the query distances are reduced, and the scopes of the effect of topological inconsistency are limited. We investigate the effect of the number of mapping replicas on query performances for DHT-based mobility support. And we find that replicating mobile nodes’ mappings in proportion to their call-mobility-rates minimizes the query delay and ensures fairness in the query load distribution. Moreover, we combine the hierarchical structure with the domain-level DHT-based (dDHT) structure, to reduce the expense of replication on update performances. We propose a hierarchical DHT and proportional replication based mobility management (HDPRMM) scheme. The numerical results show HDPRMM optimizes the query and update performances of dDHT, and achieves better fairness and network scalability than MIP and dDHT.

In mobile Internet era, the information locked in a fixed site will be released into space and time, and the flowing world will deeply change our lives anytime and anywhere. One direction of mobile Internet is large-scale development, and scalable mobility support is an important task. In typical mobile Internet architectures, a few types of mobility agents are used to maintain the mobility of a mobile node (MN): a home agent (HA) in Mobile IPv6 (MIPv6) [1], a mobility anchor point (MAP) in Hierarchical MIPv6 (HMIPv6) [2], and a local mobility anchor (LMA) in Proxy MIPv6 (PMIPv6) [3]. If a high burden of tasks is concentrated on a single mobility agent, the mobility agent may become a bottleneck node. How to distribute the network load among multiple mobility agents is an important issue. To address the scalability problem, a number of P2P-based mobility supports were proposed [4], [5], [6], [7], [8], [9], [10], [11], [12] and [13]. In our work [14], by classifying these schemes and comparing them with the traditional MIP mechanisms, we showed that the P2P-based mobility supports are highly user scalable and load balanced, but these schemes have shortcomings in query and update performances especially for a large-scale network. It is because although routing of the overlay itself is effective, there is inconsistency between logical and physical topologies, so the actual physical network performances are not necessarily efficient. Moreover, we showed that the DHT-based information spreading way and the information replication method of P2P networks have the reference value for mobility management. However, the related works only used the distributed structured information spreading way, but did not notice that using the information replication method may obtain benefits on the query performance, load balancing and data availability. In this paper, we introduce replication into DHT-based mobility management. Then all queries from any subnet can get responses as early as possible, i.e. the query distances are reduced, and the scopes of effect of topological inconsistency are limited. We construct the optimization problems with the objectives of optimizing query performances, and choose the optimal replica distribution by theoretical analysis. We show that DHT-based query is optimized when the number of mapping replicas is proportional to the MN’s call mobility rate, and such a replica distribution also ensures query load balance. In addition, to reduce the expense of replication on update performances, we combine the hierarchical structure with the dDHT management structure, and propose a hierarchical DHT and proportional replication based mobility management scheme, called HDPRMM. The numerical results show HDPRMM uses the relationship between performances and user behaviors better, and greatly enhances query performances, update performances, fairness, and scalability compared with dDHT, MIPv6 and HMIPv6. The rest of this paper is organized as follows. Section 2 describes the structure and performance problems of DHT-based mobility supports, and summarizes the related works on replication problem. The system parameters and assumptions are given in Section 3. Section 4 models the replication problem and analyzes the optimal replica distribution. In Section 5, we present the hierarchical DHT and proportional replication based mobility management scheme HDPRMM. The performance evaluation results are presented in Section 6. Conclusions are given in Section 7.

In order to provide scalable services, DHT-based distributed mobility management structures are adopted by many researchers. But in these structures, the inconsistency of logical and physical topologies limits the effective routing of overlay itself. In this paper, we introduced replication into the DHT-based management structure, to reduce the effects of inconsistency on mobility management. We constructed the optimization replication problems with the objectives of minimizing the average query delay and minimizing the average query overhead. Moreover, we analyzed the optimal replica distribution for balancing the query load, balancing the delay and overhead performances, and balancing the query and update performances. The theoretical analysis results showed the DHT-based query overhead is optimized when the number of replicas is square-root proportional to the MN’s CMR. And the DHT-based query delay is optimized when the number of replicas is linear proportional to the MN’s CMR, and such a replica distribution also ensures fairness in the query load distribution. Further, we observed that the optimal replica distribution is determined by the CMR distribution, and the entropy of View the MathML source{CMRi∑i=1NCMRi} and system capacity affect the query performances. When all MNs have the same CMRs, the query performances under these two replica distributions are the same. In addition, we showed that the method of adjusting the replica distribution to reduce the unit update overhead is impracticable. Then, to reduce the expense of replication on update performances, we combined the hierarchical structure with the domain-level DHT-based management structure, which can reduce the overlay update rate. And we proposed the hierarchical DHT and proportional replication based mobility management scheme (HDPRMM). Further, we analyzed the effects of system parameters on performances of MIPv6, HMIPv6, dDHT, DPRMM and HDPRMM by simulation. The numerical results showed that the update performances of HDPRMM are much better than MIPv6, dDHT and DPRMM, and are equivalent to that of HMIPv6. The query performances of HDPRMM are also better than dDHT due to using the proportional replication. Moreover, HDPRMM better uses the relationship between performances and user behaviors and service features, and achieves better fairness and scalability than MIP and dDHT.