Reliable Spectrum Sensing
Spectrum sensing is the task of determining whether a particular frequency band is occupied at the moment. It is an enabling primitive of the cognitive radio technology: a device can avoid stepping on the primary user only if it is able to sense the presence of that primary user. At ReCognize, we identify several spectrum sensing issues that affect the reliability of the cognitive radio network:
1. Robust individual spectrum sensing: Many prior spectrum sensing protocols are shown to be robust; however, these protocols rely on an initial survey to obtain prior knowledge of the spectrum. A misbehaving user can affect the correctness of the survey by injecting crafted signals during the survey, thereby decreasing the overall performance of the network. We propose D-HIPSS, a survey-less spectrum sensing scheme that offers similar performance to an energy detector under ideal environment.
2. False-reporting-Sybil-resilient cooperative sensing: Individual sensing, however robust it may be, suffers from the lack of spatial diversity, especially when the user is in deep fade. As such, the network can and should pool together the sensing results of many users so as to obtain higher accuracy. When pooling together results, the final sensing decision is susceptible to false-reporting-Sybil attack in which one misbehaving user masquerades as many and return incorrect individual sensing results. We seek to combat this attack by using Sybil-resilient challenges and reputation-based aggregation.
3. Reliable spectrum sensing in presence of primary emulators: It is intuitive that given the activity of all primary users, there is a corresponding benefit offered by neighboring cognitive radio networks. If a misbehaving user pretends to be a primary user, it could deprive others of the otherwise available network resource. A reactive primary emulation attacker, one that emulates as different primary users by exploiting the sensing schedule of the network, is particularly harmful. At ReCognize, we are designing protocols that enables a cognitive radio network to escape from reactive primary emulation attacks.