Handover Algorithm In TD-SCDMA/WLAN Network Based On Sarsa

Time Division Synchronous Code Division Multiple Access (TD-SCDMA) isone of important3G standards, has play a key role in telecommunication networkservice in China. Wireless local network (WLAN), at the same time, as a mostnecessary supplementary accessing network, meets the high speed demand for a lotof consumers. Convergence of this two heterogeneous accessing networks has beena must demand for today or even next generation’s communication. The mobilitymanagement specifically handover control algorithm is one of the key technology inconvergence of these two networks. Its performance will directly inflect the qualityof service (QoS) for mobile terminal users. More importantly, the energyconsumption of mobile terminal, one of the most important factors of users’ qualityof experience, has not grows stronger with Moore’s Law. Higher data transmissionspeed, more complicated data processing technology, the energy of mobile terminalrun out faster, and user’s mobile experience will get shorter and worse.The traditional handover algorithm of mobile terminal is WLAN first category,which just concerns the receive signal strength of access point in the WLAN orbasestation in the TD-SCDMA cellular network, so it causes a lot of problem, likePing-Pong handover, QoS unsatisfied, etc. in the paper, a brand new heterogeneoushandover algorithm based on Sarsa learning has been proposed, which fullyconcerns the QoS of users’ application, cost and energy consumption of mobileterminals, improves the quality of users’ experience.Firstly, we did a deep research on the features of TD-SCDMA network andWLAN, including the converge situation and coupling modes. To make the QoS ofthese two networks comparable, the QoS mapping mechanism has been given.Secondly, we carried a thorough research on energy consumption of mobileterminal, investigated every aspect which may cause energy consumption from themobile terminal devices, network types, and network state to application types. Atthe end, we figured out energy consumption control is a complex problem whichshould be optimized by a self-learning way like mobile terminal reinforcementlearning instead of artificial parameter setting.Finally, researches on the problems of single attribute judgment andunreasonable ranking attribute weights are carried on, we proposed a brand newhandover algorithm based on Sarsa learning, which concern two of most factors thatdirectly affect users’ experience, QoS and energy consumption, make a thoughtfulhandover decision according to the state of mobile terminal from Long-term benefit maximization. Simulations results show that proposed algorithm obtains a lotimprovement in network bandwidth, end to end delay and jitter, packet loss rate andmobile terminal energy consumption.