Cognitive radio (CR) has received a substantial amount of research interest because of its great potential for significantly improving spectral utilization efficiency. In a CR network, (unlicensed) secondary users are permitted to opportunistically access the licensed spectrum assigned to primary users (PUs), provided such opportunistic and secondary access to licensed spectrum causes insignificant impairments in the quality of the service of the UP. Spectrum detection is the key functionality to meet this requirement. Energy detection is an attractive method of spectrum detection in practice, mainly because of the fact that it has low implementation complexity and does not depend on any deterministic knowledge about the primary signals. Energy detection is based on the following two critical assumptions,
1) the noise power is perfectly already known; Y
2) test statistics can accurately be modeled as independent and identically distributed random variables (i.i.d.).
In a real environment, the noise power varies from time to time. This makes it difficult to estimate the noise power and incurs an inaccuracy in the modeling of the test statistics. Since the design of energy detection depends heavily on these assumptions, the question of how to design and perform detection of energy detection spectrum in a real environment is of practical importance.