A Study Of Distributed Fountain Codes Based Data Collection Techniques In Wireless Sensor Networks

As we know, wireless sensor networks (WSNs) are data-centric networks and the collection of sensor data is often the most important task of WSNs. The nodes of WSNs are usually deployed in the relatively harsh environment or restricted environment, how to ensure the reliability of data collection is a worthy subject of study.Now coding has found an increasingly wide utilization in WSNs and receives more and more attention, it also has changed the design ideas of reliable data collection protocols. Fountain codes are rateless codes and their encoding/decoding algorithm are simple, so they are very suitable to be used as the coding scheme for wireless sensor networks. When we employ the fountain codes in WSNs, it brings the system of fault tolerance and high transmission, which meet the requirement of data collection applications. Therefore, the research and application protocol design of distributed fountain codes in WSNs is meaningful and has practical value, especially LT codes and Raptor codes which are classic and widely used fountain codes.While the employ of fountain codes technology have many obvious advantages: higher data persistence and data transmission efficiency, but the sensor network scenarios are complex and diverse, often require a specific application protocol design for the corresponding scenario. According to our research and understanding, the existing fountain codes based sensor network technologies for data collection still have many problems.In this thesis, we consider sensor networks in a disaster scenario; we focus the collection of sensor data and made the following research work:(1) We propose a data collection protocol called Regulative Growth Codes which motivated by the performance degradation of GC in sparse networks, RGC use a random sequencing strategy to improve data exchange and collection efficiency data persistence and data network for sparse liquidity Limited, since the node detection algorithm is proposed to reduce unnecessary data exchange, effectively reducing the total network data traffic.(2) Based on the latest distributed packet-centric encoding algorithm, we propose a novel distributed implementation of Raptor codes, avoid the pre-encoded overhead in the traditional distributed Raptor codes. We implement the protocol in simulation and give the performance evaluation.