Research On Differentiated Service Mechanism For IEEE802.15.4

Low-rate wireless personal area network(LR-WPAN) is a new short-range wirelesscommunication technology to provide low-rate,low-cost, low-complexity and low-powerwireless data transmission services, which has been widely used in different areas includinghome-automation, medical-monitoring and military applications. In order to meet differentengineering applications and to maintain compatibility, IEEE Standard Association(IEEE-SA)established a working group TG4to define a standard called IEEE802.15.4which specifiesthe physical layer(PHY) and media access control layer(MAC). With its low-power physicallayer and stable MAC layer,the standard has received extensive attention and is a basis forZigBee, WirelessHART and6LoWPAN specifications, each of which further extends thestandard by developing the upper layers which are not defined in IEEE802.15.4. Howerver,different data packets may have different importance depending on the informationcontained in WSN applications.Although IEEE802.15.4provide GTS mechanism fortime-critical events, a device must send a appropriate request before using GTS and onlyseven GTS slots can be achieved. Therefore, better support for802.15.4networkdifferentiated services should be provided.To provide better support in differentiated service for IEEE802.15.4, there exist twosolutions. First, a lot of researches mainly focused on increasing the utilization of GTS byimproving the GTS allocation mechanism, which was very complex to implement in generalhowever. On the other hand, some researches applied different CSMA/CA parameters (suchas CW, BE, etc) to different service from the perspective of CSMA/CA algorithm, thusgiving the high-priority service a better chance to access the channel, which was simple butdid not consider the effects of duty cycle in coordinator. In this paper, a new differentiatedservice strategy has been proposed on the basis of priority-based strategy from theend-device and the coordinator respectively, which not only provided better service forhigh-priority serivce but also adjusted ducy-cycle according to different applications tomaximize network energy savings. In terms of end-device, the CSMA/CA algorithm andeffects of different parameters on it has been intensively studied. BCS (Backoff Counter Scheme) backoff strategy was added to original priority-based differentiated service strategyto sovle the service differentialion problem under higher traffic condition. Meanwhile interms of coordinator, in order to meet the perforcement requirements of high-priority servicein different applications and maximize network energy savings, the impact of differentduty-cycle setting on packet’s latency, packet delivery ratio and network remaining energyhas been first investigated on this paper by using NS2simulator.And based on the simulationresult a reinforcement learning based duty-cycle adjustment algorithm was proposed to“self-learning” an optimal duty-cycle and can track environmental changes thus making aadjustment on duty-clcye setting.To verify the quality of the proposed algorithms, all algorithms have been realized inNS2simulator. For BCS, end-to-end delay and packet delivery ratio of two different priorityservices under different traffic conditions were simulated,which showed that the BCSstrategy provided better service quality for high-priority serivce compared to original onewhen traffic is higher. While for the Reinforcement learning, this paper not only simulatedthe learing effect of reinforcement leanring algorithm under three different applicationrequirements but also when the traffic condition was dynamically changed.Simulationresulsts showed that the proposed algorithm can meet the perforcement requirements ofhigh-priority services for different environments and adjust the duty-cycle according totraffic changes in network, which showed a strong environmental adaptability.