The entropy of wireless statistics

Supply a non-deterministic random number generator with a fresh source of entropy is a challenge for wireless mobile devices with limited resources. Different works propose to use environmental measures or physical noise as entropy sources for WSN. In this paper, we aim to analyze the entropy generated by the wireless channel using the IEEE 802.15.4 standard. To achieve this goal, we focus on erroneous packets, RSSI (Received Signal Strength Indicator) and LQI (Link Quality Indicator) used to reflect respectively the signal's power and the quality of the channel. No additional hardware is needed to gather data from these sources, which make them more interesting for WSN. In order to estimate the entropy of such sources, we propose a new analytical framework using the pessimistic entropy estimators recommended by the NIST. We have lead several experiments to evaluate the entropy collected under various conditions, and to study the impact of some attacks (e.g. eavesdropping or jamming attacks) on these entropy sources. Analysis shows that the LQI indicator and the erroneous packets are the most interesting entropy sources. These data are random and independent from one node to another, and seems to resist eavesdropping attacks.