Study On Packet Scheduling Algorithm In Heterogeneous Wireless Multimedia Sensor Networks

Wireless multimedia sensor network (WMSN) is a new type of WSN with video,audio and image functions. Heterogeneous Wireless Multimedia Sensor Networksincludes scalar sensor nodes, image sensor nodes, vedio sensor nodes and many otherkinds of sensor nodes. It has better performance, better scalability, wider coveragearea and higher reliability.A WMSN often consists of a large amount of sensors, which constantly generateall kinds of traffic. The packets of these event-driven data belong to different levels ofemergencies, and have their own valid periods. Thus, the packets from sensors arerequired to be delivered to the sink before their expiration deadlines which are theirquality-of-service (QoS) requirements. There are two challenges for WMSNs tosupport a huge amount of data that has diverse deadlines. First, the sensor nodes havevery limited resources of energy supply, bandwidth, memory and processingcapability due to the physically small size of sensors. The limited resources require afair and efficient utilization. Second, the packets with the same level of emergencyhave the same deadline requirement. The packets from the sensors which are far fromthe sink have to traverse more hops to reach the sink than the packets from the sensorswhich are close to the sink. Thus the delay requirement of remote event report packetsat every intermediate node is more stringent than that of nearby event report packets.Packet scheduling is an important scheme to address the QoS requirement issuesin WMSNs. Generally, packet scheduling assigns priorities to traffic flows with QoSconsiderations. Then packets are queued and scheduled according to the assignedpriorities, which are related to the effective periods of event reports. If a packet cannotbe sent to the sink in time, it will be discarded, resulting in high packet drop rate.With the limited resource in WMSNs, the packets of the most emergent event, whichhave the closest deadline, should be serviced with the highest priority. On the otherhand, the geographic distribution of sensors also has impact on the schedulingpriorities. The delay requirement of the packets from the remote sensors is morestringent than that of nearby event report packets, so they should be serviced withhigher priority. For these considerations, packet scheduling should assign priorityaccording to a packet’s deadline and the distance from its source to the sink. Most packet scheduling algorithms do not meet this requirement. So thedissertation proposes Wireless Differentiated Queuing Services (WDQS) algorithmwhich is suitable for WMSNs. WDQS queues packets at intermediate node by theirlatest depatrue time (LDT). The smaller the LDT, the higher the priority. A packet’sLDT is determined by subtracting the known delay which it has already experiencedwhen arriving at the intermediate node and the unknown delay which it willexperience after leaving the intermediate node from its lifetime which can be setartificially. The difference between its lifetime and the delay which it has alreadyexperienced is its deadline. The known delay and unknown delay correspond to thedistance from its source to the sink since our study is based on geographic routing.Thus the packet’s priority is higher if its deadline is shorter and the distance from itssource to the destination is longer. But how to estimate the unknown delay to thedestination and how to set the packet’s lifetime is hard to be resolved. Ourcontributions include:1. Packet scheduling scheme: The scheme assigns assign priority according to apacket’s deadline and the distance from its source to the sink.2. Effective delay estimation approach: We propose a effective delay estimationfor the remaining journey of a packet to implement WDQS by exploiting the uniquecharacteristics of WMSNs. Our approach introduces no extra message overhead andenergy consumption into WMSNs.3. Packet lifetime setting: We deduce the expression of packet lifetime setting insingle data source ideal conditions, multi-data sources ideal conditons and unidealconditions. We get the general packet lifetime setting conditions. The networkperformance is anlysised with the conditions is satisfied and unsatisfied respectively.Research from simple to complex discusses all kinds of situations.4. Simulation: There are two kinds of applications of WMSNs: periodic datacollection application and bursty data driven application. CBR and passion processare suitable for periodic data and FBM process is suitable for bursty data. We deducethree concrete condition expressions from general conditions expressions with thearrival process is CBR, passion and FBM respectively. These three expressions areverified by simulations. The results of simulations show they are reasonable. So thegeneral conditions are reasonable and fit for all kinds of WMSNs applications.5. Realization: There are three kinds of nodes in heterogeneous WirelessMultimedia Sensor Networks: sink node, source node, and intermediate node. WDQS parameters are related to the status of the network. Manual setting is difficult. Westudy the algorithm implementation on the three kinds of nodes. The parameters canbe set automaticly. We make WDQS to be realized possible.