Received Signal Strength Based The Key Generation Matching Analysis In Wireless Sensor Networks

Wireless sensor networks(WSNs), a new information acquisition and information management technology, has been widely applied in many fields. Particularly, its security issue has also become one of tips in wireless sensor network research. In order to guarantee the security of WSNs, one of the most important mechanisms is to use a valid key in communication. Recently, key generation method based on the physical characteristics of the wireless channel, has verified to be a valid alternative security design compared with the conventional ones. Many scholars, at home or abroad, have also done much research. However, some problems expose in the exciting key generation methods, such as a bad matching and a low entropy.This paper presents a new adaptive secret key extraction protocol by improving the existing methods based on deeply understanding the definition, characteristic, the channel characteristics in WSNs and recognizing the security and challenge in this field. The protocol generates secret key through Received Signal Strength(RSS) on the wireless channel between the two devices and increases the bit mismatch rate and entropy effectively. Three steps are devised in this method. a)Further pre-treat on the recovering mean RSS using the new designed protocol, that is, removes the coupled RSS with a big difference value to reduce the effect of environmental noise on the Sampling RSS values; b) Propose two quantization methods in order to improve bit mismatch rate, mean uniform quantization and mean non-uniform quantization, both of which take the same quantitative thresholds; c) Further process the result after quantization and remove the RSS value in the middle area in order to obtain the same secret key.This paper uses crossbow’s MICAz and MIB520 CA as the hardware platform and the testing algorithm is written in nes C executed on Tiny OS. We use sensors to collect data from different real-world wireless sensor networks, and process the data through software simulation protocol. By setting the variable parameters in the protocol to different values, the paper compares and analyzes the performances of our proposed protocol with the existing schemes on bit mismatch rate, secret bit rate and entropy. This paper analyzes the impact of different variable parameters on the performances of the protocol. Our measurements show that our scheme, in comparison to the existing ones that we evaluate, performs better in terms of generating high entropy bits at a low bit mismatch rate. In addition, removing the RSS values which have a big difference between two sides has no impact on the entropy. But with the D_value becoming smaller, the secret bit rate decreased rapidly. The increasement of the block size w used to compute μ value can reduce the bit mismatch rate, but it has no effect on entropy and secret bit rate when the two devices recover the RSS by mean value. With the increasing of the quantization level q_level, the bit mismatch rate also increase, that is same to the entropy and secret bit rate.