A strategy to improve sensing accuracy of energy detection for distributed spectrum management systems

Spectrum sensing has a key role in the dynamic spectrum access' paradigm. Spectrum sensing must be carried out to identify transmissions of licensed users, and therefore, avoid harmful interference on them. In addition, it enables the discovery of opportunities that allow communication between the cognitive radio devices to take place. Energy detection is a spectrum sensing technique with some important advantages: simplicity, low response time and low computational cost. However, this technique has very low accuracy in comparison to other approaches. This paper presents and evaluates a proposal to improve the accuracy of spectrum sensing based on energy detection. Our proposal employs hidden Markov model (HMM) to distinguish signals from licensed devices in scenarios where their SNR values are comparable to the ones from cognitive devices. Performance evaluation using ns-3 simulator have shown that our proposal improves the discovery of transmission opportunities.