Research On Data Delivery And Application In Opportunistic Networks

A wireless ad hoc network is a decentralized, self-organizing and multi-hopwireless network, which does not rely on any pre-existing infrastructure. It has theadvantages of high flexibility and swift deployment, and thus has many practicalapplications, especially in the scenario without infrastructure or the infrastructure isdestroyed owing to the earthquake or other uncontrollable factors. In many applications,such as the one which exploiting the sensors carried by animals to collect the migrationdata and the one which taking advantage of the networks formed by the movingvehicles to achieve accident warning, the network is intermittent owing to the mobilityof node, the scarcity of the network or the attenuation of the communication capabilitywith the distance. That means it might not exist an end to end path among differentnodes. To achieve effective communication in the above scenario, the opportunisticnetworks are proposed. In opportunistic networks, data transmission needs to be carriedout in a carry and forward way using the communication opportunities brought by nodemobility. As a special case of ad hoc networks, the opportunistic networks do notrequire a connected path among the source node and destination node, which satisfy thedemand of practical applications very well, and thus have a great significance on theprocess of realizing ubiquitous computing.Data delivery is an important research topic in opportunistic networks. However,the inherent characteristics of opportunistic networks, such as the node mobility andintermittent connectivity, make data delivery over it a challenging issue. The nodemobility is a double-edged sword. While extending the data delivery in both time andspace, it also makes data delivery extremely complicated. Moreover, the opportunisticnetworks have a broad range of applications. Different applications have differentcharacteristics, which makes the data delivery closely related to the specific scenario. Inview of the aforementioned fact, this dissertation focuses on two typical types ofopportunistic networks, Delay Tolerant Mobile Sensor Networks (DTMSN) andVehicular Ad Hoc Networks (VANETs), and investigates the data delivery issue in theseapplications. After analyzing the characteristics of these two types of networks thoroughly and systematically, this dissertation makes an in-depth and concrete study ondata delivery in the corresponding scenarios. Apart from the data delivery, applicationdevelopment is another key research topic in opportunistic networks. As they extend thedata delivery both in time and space, opportunistic networks expand the applicationscope of traditional ad hoc networks greatly. Thus, it has great practical meaning toconduct in-depth research on the application of opportunistic networks. This dissertationconducts a thorough analysis of the characteristics of VANETs. Motivated by the factthat it is difficult to find an available parking space in crowded urban areas, it proposesa novel parking guidance scheme, which doesn’t rely on any infrastructure and couldprovide efficient guidance for vehicle users.The major contribution in this dissertation is as follows:1. The characteristics of human mobility are analyzed, and a novel delay-basedrouting protocol (DRP) is proposed for human-oriented DTMSN. In DRP, a sensor nodecalculates the estimated data delivery delay and takes it as a measure of the deliverycapacity, the smaller the estimated delay, the higher the delivery capacity of a sensornode. When two nodes meet, data messages are forwarded to the one with smallerestimated delay. To minimize transmission overhead, DRP employs the message rankand survival time to decide message’s transmission or dropping. The proposed schemeis evaluated in a real human moving scenario provided by MIT Reality datasets, whichcollected the traces of100individuals of MIT over the course of9months. Simulationresults have shown that the proposed DRP routing protocol achieves a betterperformance than some other DTMSN data delivery approaches.2. In view of the unique characteristics of VANETs, e.g., high mobility of vehiclenodes, intermittent connectivity, and rapidly dynamic topology, a parked-vehicleassisted data dissemination (PADD) paradigm is proposed for Vehicular Ad HocNetworks. In PADD, data to be disseminated is maintained at the roadside parkedvehicle, which continuously provides services for the vehicles passing by. Thisdissertation analyzes the challenging issues in parked vehicle assisted datadissemination and proposes one possible solution for each issue. To be specific, theparked vehicles in the target area were managed on the basis of parking cluster; arouting scheme was designed for data forwarding from the data source to the selectedparking clusters; the pub/sub scheme was adopted to achieve data dissemination within one hop of the parking cluster. Theoretical results illustrate the effectiveness of ourapproach, and simulation results based on a real city map and realistic traffic situationsshow that the proposed data dissemination paradigm achieves a higher delivery ratiowith lower network load and reasonable transmission delay.3. Motivated by the fact that it is difficult to find an available parking space incrowded urban areas, an intelligent parking guidance scheme IPARK is provided forurban VANETs, which taps into the unused resources (e.g., wireless device,rechargeable battery, storage capability) offered by parked vehicles to perform parkingguidance. In IPARK, the cluster formed by parked vehicles generates the parking lotmap automatically, monitors the occupancy status of each parking space in real time andprovides assistance for vehicles searching for parking spaces. Overall, IPARK couldprovide the drivers with two kinds of services: parking availability informationdissemination outside the parking lot and real-time parking navigation inside theparking lot. This dissertation proposes an efficient architecture for IPARK andinvestigates the challenging issues in realizing parking guidance over this architecture,e.g., how to manage the parked vehicles and how to generate the parking lot mapautomatically. Finally, this dissertation investigates IPARK through realisticexperiments and simulation. The numerical results obtained verify that IPARK achieveseffective parking guidance in VANETs.