AN OPTIMAL FRAMEWORK FOR INTRUSION PREVENTION IN WIRELESS SENSOR NETWORKS

Abstract

The field of wireless sensor networks (WSNs) combines sensing, computation, and communication into a single tiny device called a sensor. Sensors are equipped with RF radio, processor, memory and hardware. They are also battery powered and therefore have severe energy, bandwidths and memory constraints, and low computational capability. Communication over WSNs is still known to be attack-prone because the constraints of sensors hinder the development of secure modern cryptographic solutions. The Elliptic Curve Cryptography (ECC) technique and the Rivest Shamir Adleman (RSA) algorithm are the two most popular public key cryptographic schemes deployed over wireless networks. The effectiveness of the ECC technique over RSA has been demonstrated in this research. ECC with very large key sizes are thought to be computationally expensive; it is possible to use smaller primes, or smaller finite fields, with elliptic curves and achieve a level of security comparable to that for much larger integer mod n. Measurements have been made to prove that ECC algorithms can be executed within the memory limits of sensor nodes. An enhanced ECC scheme with collision resistant hash functions was been implemented on the J-Sim platform and measurements were conducted to determine possible improvement in network lifetime. Security simulations were been carried out. It was observed that the developed scheme consumes much less energy while still providing adequate level of security in comparison with other security models in existence. Results for computation time, energy consumption for selected radio models of sensor nodes are presented. The effect of selected variables on sensor node lifetime is described and analyzed. A comparison with other models for security and energy consumption is made and suggestions for future work are presented