International Journal of Computer Applications (0975 – 8887) Volume 91 – No.10, April 2014 8 Energy Efficient Wireless Sensor Network using Genetic Algorithm based Association Rules T. Abirami, Ph. D Assistant Professor (SRG), Department of Information Technology Kongu Engineering College Perundurai, Erode, Tamilnadu, India P. Priakanth, Ph. D Professor, Department of Computer Application Kongu Engineering College Perundurai, Erode, Tamilnadu, India ABSTRACT Wireless Sensor Networks (WSN) usually contains thousands or hundreds of sensors which are randomly deployed. Sensors are powered by battery, which is an important issue in sensor networks, since routing consumes a lot of energy. Such nodes are deployed in thousands to form a network with capacity to report to a data collection sink (base station). An efficient routing scheme in sensor network is also important. Networking unattended sensor nodes are expected to have significant impact on the efficiency of many military and civil applications such as combat field surveillance, security and disaster management. Genetic algorithm (GA) based data aggregation trees are used where the sensors receive data from neighboring nodes, aggregate the incoming data packets, and forward the aggregated data to a suitable neighbor. GA is used to create energy efficient data aggregation trees. In this work, the amount of data sent to sink is reduced using association rule mining and in turn to further reduce the energy consumption of the network; optimal routes are chosen to transmit data to the sink based on energy consumption. The proposed method is able to discover the association rules to make predictive analysis on node failure, asymmetric links. The rules found form the basis for coding solutions in the proposed genetic algorithm. GA is applied to generate balanced and energy efficient data aggregation spanning trees for wireless sensor networks. E-Span, which is an energy-aware spanning tree algorithm and Lifetime-Preserving Tree (LPT) are used to create data aggregation trees. The proposed GA extends network lifetime. Keywords Wireless sensor networks, genetic algorithm, energy efficient, data aggregation Trees, Association Rules 1. INTRODUCTION The challenge of improving sensor energy consumption is a key issue in wireless sensor networks (WSNs). This is due to the fact that sensor network lifetime is directly related to operational lifetime. If sensors can operate for longer periods of time and the frequency of node failures can be reduced, the reliability and adaptability of the sensor network will consequently improve. Data transmission is very costly in terms of energy. Sensor which collect data hand them over to the sink which is followed by offline data analyses to extract patterns. The existence of a large communication overhead affects sensor network performance negatively. This large overhead becomes a hurdle for the deployment of long term large scale sensor networks. Association mining is used to discover frequent patterns in the data. As the association mining is applied in-network, Patterns and not the raw data streams are forwarded to the sink when association mining is applied to the network which thereby reduces communication overhead significantly. Sensor nodes that are used to form a sensor network are normally operated by a small battery which has small amount of energy. Therefore, in wireless sensor networks reducing energy consumption of each sensor node is one of the prominent issues to address in the network lifetime, since wireless communications consume significant amount of battery power, sensor nodes should be energy efficient in transmitting data. Protocols can reduce transmitted power in two ways. First where nodes can emit to short distances such as data sinks or cluster nodes. The cluster node can then send the data over a larger distance preserving the power of the smaller nodes. The second is by reducing the number of bits (amount of data) sent across the wireless network. In this work, the amount of data sent to sink is decreased using association rule mining and to further reduce the energy consumption of the network, optimal routes are chosen to transmit data to the sink based on energy consumption. A genetic algorithm (GA) is implemented to generate balanced and energy efficient data aggregation spanning trees for wireless sensor networks. In a data gathering round, a single best tree consumes lowest energy from all nodes but assigns more load to some sensors. As a result, the energy resources of heavily loaded nodes will be depleted earlier than others. The proposed GA extends network lifetime. 2. MATERIAL AND METHODS 2.1 Dataset To evaluate the proposed methods, the intel lab sensor dataset is used [5]. This dataset contains data collected from 54 sensors installed in the Intel Berkeley Research lab. The sensors installed were Mica2Dot sensors with weather boards. The data collected were timestamped. The data collected consists of topology information, humidity, temperature, light and voltage value. The data was collected once every 31 seconds. The dataset includes a log of about 2.3 million readings from the sensors. The data is represented as shown in Fig 1.