A nonparametric sequential kolmogorov-smirnov test for transmit opportunity detection at the MAC layer

Efficient and reliable detection of transmission opportunity is the key enabler of cognitive radio networks. This paperintroduces a novel transmission opportunity sensing paradigm which operates at the MAC layer of the primary network in contrast to most previous work. We propose a non-orthogonal overlay architecture based on detecting changes in the probability density distribution of primary network packet statistics. Specifically, we develop a novel sequential version of the Kolmogorv-Smirnov goodness-of-fit test which allows the secondary network to operate subject to an interference constraint that ensures a given QOS in the primary network nodes. We provide an efficient implementation of the test on an experimental testbed and demonstrate its utility and viability under field testing on an IEEE 802.11 WLAN. Compared with results reported so far in the literature, our experiments achieve a significantly improved 50%–75% utilization of hitherto unused transmit opportunities while limiting the probability of interference to ≪ 0.02. The average detection delay is 140 ms to 290 ms and compares favorably with the traditional physical layer based approaches.