Abstract
Abstract
This master thesis presents the effect that security has on a sensor network. A wireless sensor network is composed of a magnitude of sensors that together form a sensor network through self- organizing and self- configuring capabilities. A sensor node is small unit that functions as a transformer which converts physical signals into electrical signals. The sensor unit itself is composed of a sensor, a processor, a transceiver and a power supply like a battery. This battery is a small sized battery, so it is important to optimize all the components and software, so that the sensor nodes consume as little power as possible. Introducing security in a system which has so many restrictions as a sensor network is very challenging. All these challenges rules out the use of asymmetric encryption techniques, which make symmetric encryption techniques more relevant for a sensor network.
This report gives an overall description of the sensor network technology. We present security techniques that can be suitable for a sensor network. Additionally we look at the degree of security that we can obtain with the different security techniques. It is important that security is chosen carefully, so that the cost associated the particular security techniques don’t drastically degrade the lifetime of a sensor node. Then we are going to illustrate the problems that arise when security is to be introduced in a sensor network. Further we have studied the overhead associated with known symmetric techniques, and also how this overhead effects the packet growth, throughput, transmission time, and how this effects the power consumption. The power consumption has been calculated through several scenarios, where we investigated how the lifetime of the sensor nodes will degrade when different security overhead is introduced into the data packet, and also how the node density can effect the power consumption in the sensor network. We have also tested how a sensor network operates at different load, and we have calculated metrics like power consumption, end to end delay, and average throughput in the sensor network. All these results have been obtain through simulations with the use of NS-2 network simulator and Sensorsim, and the results shows that our suggestion for a security framework can be implemented in a sensor network.