S-MAC Protocol And Reorgnization Algorithm In Wireless Sensor Network

Wireless Sensor Network (WSN) is usually composed of a large number of sensor nodes which are normally limited in power, computational capacities, and memory. The position of sensor node allows random deployment in inaccessible terrains such as battlefields, volcanoes and Polar Regions or disaster relief operations. For the limits of single nodes'transmission range, the communication must be in multi-hops manner. This means that protocols and algorithms for sensor networks must be designed in self-organizing and distributed manners. Within all the characters of WSN, energy consumption is almost the most important problem. Since sensor nodes are often deployed in hazard environments, it is very difficult to change or recharge batteries. So the most important object of the research is to extend the life time of WSN. Partitioning is a calamity for the network, because it will make the sub-networks that disconnected to sink nodes can't send data to a base station any more.The main topic of this thesis is discussing the methods of preventing the network from partitioning in order to postpone the finality the whole network. The works of this thesis are as follows:1. According to our experience, we summarized that these following factors are the reasons of network partitioning:(1) Some nodes are deployed in areas of high probability of occurrence, these nodes'batteries may deplete faster than others because they undertake more transmission and sensing tasks.(2) Since sensor nodes are deployed by scattering manner, its densities of areas are different from the beginning, so nodes in sparse area will undertake more tasks.(3) The characters of protocols may also cause power distribution of network unbalance, such as the virtual cluster border nodes problem of in S-MAC protocol. (4) Since the hazed environment, external factors may also cause the die of sensor nodes.2. We proposed a distributed algorithm in detect critical node for wireless sensor networks.Seriously, some of the depleted nodes may play a role of critical nodes whose removal will separate the network into two or more pieces.Traditionally, DFS (depth first search algorithm) was widely used to detect critical nodes and critical links, but DFS is a centralized algorithm, it requires that a computer should be aware of the global topology of the network. Although it can be also implemented in globalize distributed manner, which is achieved by exchanging of routing tables or position information among each nodes, than a node collect the information of global topology and compute it. This method is inefficient in practice and involves a quadratic (in number of nodes) communication overhead in order to update topology information. This mean it does not belongs wireless sensor networks because of the limited capability of sensor nodes.In this article, we proposed an algorithm to detect critical node, in which a sensor node needn't know the information of network topology and position of other nodes, a node that has the information of its neighboring list can predict whether it is a critical node. The algorithm is triggered when a node N finding its power lower than a threshold. After its end, it is not only determined whether a node N is a critical node, but also found out how many sub-networks there will be, and any other nodes that had joined in the algorithm are aware of which sub-networks it belongs to after the critical node's died.3. In order to prevent the network from partitioned, we proposed a scheme to reorganize the network.Once a node identifies itself as a critical node, it must take actions to prevent the networks from partitioning. It triggers the network reorganization process which can prevent the network from partitioning. Our scheme is let node N produce a power-level message and broadcast in k-hop area, by maintaining the sub-network information which is collected in the determining process, any node that receive this message will magnify its communication radius in order to probe new neighbors from different sub-networks.3. Designed and implemented the above algorithm and scheme based on S-MAC protocol. S-MAC is a contention-based MAC protocol explicitly designed forsensor networks. We analyzed the main characters of S-MAC protocol and find the interfaces with that we can integrate our algorithm onto S-MAC.4. Our algorithm and scheme is simulated in NS-2 platform which is widely recognized in academe.We simulated the execution time and power consumption, and the influence of latency when the network is reorganized in k-hop area. The simulation results show that it works well in preventing networks from partitioning and connectivity maintenance and reduced latency when the network is reorganized in a local area.Although it is explicitly designed for sensor network, this algorithm also can be used universally for other multi-hop networks which are in large scope and limited in single nodes. Its main characters are as follows:(1) It is not a real time algorithm, it not belong the networks of users as nodes, but just adapt to the long duty character of WSN.(2) Nodes only need to receive/transmit massages and a little internal execution. This is in agreement with the designing principle of WSN, which is a powerful network is composed of a great number of mini nodes.In conclusion, the works of this thesis availably and effectively prevented wireless sensor networks from partitioning, improved the robustness, and prolonged the life time of the network.