Volume 5 - Issue 2
Robust Detection of Rogue Signals in Cooperative Spectrum Sensing
- David Jackson
Virginia Commonwealth University, Richmond, Virginia 23284, United States
jacksonds3@vcu.edu
- Wanyu Zang
Virginia Commonwealth University, Richmond, Virginia 23284, United States
wzang@vcu.edu
- Qijun Gu
Texas State University, San Marcos, Texas 78666, United States
qijun@txstate.edu
- Meng Yu
Virginia Commonwealth University, Richmond, Virginia 23284, United States
myu@vcu.edu
Keywords: Cognitive Radio Network, Cooperative Spectrum Sensing, Rogue Signal, Trust Model
Abstract
Cognitive radio networks can sense, detect, and monitor their surrounding radio frequency conditions
including the interference and availability of a broad range of channels, followed by choosing the best
possible channel for a given task. This is called Dynamic Spectrum Access (DSA) and it is a key
characteristic of cognitive radios that enable them to operate on unused licensed channels.
However, increased flexibility and convenience often leads to greater security attack vectors.
Since cognitive radio networks perceive their surrounding environment through the physical layer
in order to make decisions, they become vulnerable to rogue signals. Depending on how rogue
signals are used, they can achieve Primary User Emulation, Sensory Manipulation, or Rogue Signal
Framing (RSF) attacks. Our work focuses on accurately detecting rogue signals to mitigate the
damage of RSF attacks on trust-based Cooperative Spectrum Sensing (CSS) protocols. We devised
a community-detection clustering algorithm to distinguish between malicious/malfunctioning sensors
and well-behaved sensors affected by rogue signals. Our rogue signal detection improves upon
previous work through the use of dynamic clustering methods on a group of sensors based on the
network’s size and density over a particular region. This gives the advantage of a one-size-fits-all solution
when it comes to handling networks that are sparse, dense, and disproportionate. Additionally,
we ran extensive tests that demonstrated an upward of 6% to 40% improvement, depending on the
scenario parameters, in detecting rogue signals.