Comparison Of Random Key Predistribution Schemes In Wireless Sensor Networks Based On Average Distance

Sensor nodes carry limited, generally irreplaceable power sources and as a result the lifetime of a sensor network depends on the lifetime of the power units of the nodes, the drain of which is mainly caused by the data communication, including both transmission and reception, out of all the three domains involved with power consumption, namely sensing, communication, and data processing. In random key predistribution schemes, a set of keys are randomly selected from the key pool and preloaded into each sensor node's memory before deployment for secure communication purposes. After deployment and key setup is complete, a secure graph (wireless network) of connected links is formed as a result of the pre-distributed shared keys between the nodes. It is worthwhile therefore, to determine the average distance that a message takes to be communicated or relayed from the source to the destination in such a network in order to conserve the limited node's energy and take action in the shortest time possible on the sensed data for the sensor nodes that are involved in areas that require real time action like in military zones.Therefore, this thesis gives a new approach of comparing random key predistribution schemes in WSN (Wireless Sensor Networks) based on average distance. Average distance is the mean of distances or links over all pairs of nodes. The average distance is calculated from the network formed as a result of the key shared between the nodes as per the random key predistribution scheme applied. The evaluation criterion for this thesis is that the lower the average distances the better the scheme. The advantages of low average distance in wireless networks formed as a result of random key predistribution schemes are discussed, prominent of which is the effective use of network resources by conserving energy. The schemes considered are the basic scheme, q-composite, Random-pair-wise, and the Closest pair-wise. Experiments showed that the closest pair-wise scheme performed better while the random pair-wise scheme performed poorly with average distance. The basic scheme performed also better than the q-composite scheme. It was also found out that by increasing the node degree reduces the hop count and hence reducing the average distance. The other factors that were found out to contribute to the value of the average distance were the key ring size, transmission range of the nodes, deployment pattern and the network size (node density). Through the simulation, it was also observed that while the random pair-wise scheme performed poorly with average distance, the number of nodes without a shared key with their neighbors within a nodes wireless communication range after key discovery, decreased considerably with increase in the key ring size. Accordingly, an enhancement was proposed for the random pair-wise scheme whereby a unique key is placed in the node's key ring space to be used during key establishment. Experiment showed that this trivial enhancement had a remarkable effect on the average distance of the random pair-wise scheme. From the simulation and analysis, it was found out that while the main emphasis of evaluation of random key predistribution schemes have been hinged on the resilience of the resultant network, the average distance should also be considered for the longevity of the network life by conserving its energy with low average distance.