The Research On Dynamic Data Gathering Technology For Delay Tolerant Sensor Network

Wireless sensor networks have wide-range potential applications. They have beenconsidered as one of the most important techniques of the twenty-first century, and haveattracted a great deal of research attentions in recent years.Data gathering is a major function of wireless sensor networks. The traditionaldata information gathering approach usually relies on a large number of denselydeployed sensor nodes with short range radio to form a well connected end-to-endnetwork. Sensors in the network collaborate together to collect the target data andtransmit them to the sink nodes. This approach, however, may not work effectively inthe scenarios with extremely low and intermittent connectivity due to sensor node'srandomly deployment, sparse network density, sensor energy exhaustion or sensor nodemobility and so on. Apparently, the network may not always connective in theaforementioned scenarios and the characteristic that the connectivity of the networkvaries over time is called intermittent connectivity. Intermittent connectivity is a maincharacteristic of Delay Tolerant Network (DTN), while a sensor network where sensornodes are intermittent connective to each other is called Delay Tolerant Sensor Network(DTSN). Furthermore, DTSN with mobile sensors is called Delay Tolerant MobileSensor Network (DTMSN).Since the characteristic of intermittent connectivity, the traditional data gatheringtechnology of wireless sensor networks can not be applied in DTSN. Based on asystematical summary of relevant works on DTSN data delivery technology, thisdissertation studies the following issues: 1) How to transmit data messages to one ormore sink nodes in network with sensor node's intermittent connectivity as DTMSN? 2)How to deliver events according to their types in DTSN? In other words, how to deliverevents to all their interested subscribers respectively in DTSN? 3) How to deliver eventsto their interested subscribers with high average data delivery ratio and low average datadelivery overhead? The innovative research results of this dissertation are as follows.1. This dissertation proposes a new data gathering method-SRAD (SelectiveReplication-based Adaptive Data Delivery Scheme). SRAD achieves a good network performance by replicating messages selectively to nodes that has the higher probabilityof meeting the sink. SRAD consists of two key components for data transmission andqueue management, respectively. The former makes decisions on when and where totransmit data messages according to the node delivery probability in Random waypointmobility model. The latter employs the message survival time to decide message'stransmission and dropping for minimizing transmission overhead. Simulation resultshave shown that the proposed SRAD data delivery scheme not only achieves arelatively long network lifetime but also the higher message delivery ratio with thelower transmission overhead and data delivery delay than some other previous solutionsdesigned for DTMSN.2. This dissertation gives a community based publish/subscribe system model forDTSN. In this model, sensors in the network are divided into several independentcommunities.Then a community based event transmitting protocol (CET) inpublish/subscribe system tailored for DTSN is presented. The core idea of CET is thatevent transmitting is based on communities which are formed by sensors in the networkaccording to their connectivity. CET consists of two key components for datatransmission and queue management, respectively. In data transmitting, not only eventsare transmitted to mobile subscribers as much as possible but also some events insubscribers are retransmitted to sensors in community, for enhancing the data deliveryratio. The queue management employs both the event survival time and successfullydelivery time to decide whether the event should be transmitted or dropped forminimizing the transmission overhead. Simulation results have shown that the proposedCET achieves a higher event delivery ratio with the lower transmission overhead andevent delivery delay than DG (direct gathering protocol).3. This dissertation exploits distributed group detection from the human trace andproposes GAPC, a group-based adaptive publish/subscribe communication protocolover detected groups for DTMSN. The core idea of GAPC is that publish/subscribecommunication is based on groups which are classified based on number of contactsand contact duration of people in the monitoring area. The sensor with highestremaining energy in each group is selected as group head node then. GAPC consists oftwo components for event transmission and queue management. In event transmission,not only events in the group are transmitted to mobile subscribers but also some events in a subscriber are retransmitted to sensors that beyond the group of the subscriberreside in, for enhancing the data delivery ratio. To avoid flooding, the maximumretransmitting time is prescribed. The queue management employs the event survivaltime to decide event's transmission and dropping for minimizing transmission overhead.The effectiveness of GAPC is demonstrated through comprehensive simulation studies,...