# Introduction

The idea of Cognitive Radio (CR) is to allow unlicensed users (CR users) to use spectrum holes that are not occupied by licensed users in order to maximize spectrum utilization. The first task for the CR users is to detect the existence of a spectrum hole or spectrum opportunity as depicted in the figure and it is called spectrum sensing. The problem of spectrum sensing in cognitive radio can be viewed as detecting the presence or the absence of primary users activity. The essence of spectrum sensing is a binary hypothesis problem between the presence and the absence of the licensed user's signal transmission in a certain band. There are mainly two approaches: local spectrum sensing, which is a single user approach and collaborative spectrum sensing, which takes the advantage of spatial diversity by allowing collaboration between CR users.

# Research Interest

Designing a detector for spectrum sensing in cognitive radio networks raises the following challenges:

• Multipath fading. It is an inherent problem in wireless communication networks.
• The hidden terminal problem. It occurs when a CR user is shadowed or outside the coverage area of the licensed transmitter as depicted in Fig.2.
• Noise uncertainty and outliers. They occur in real world implementations and degrade the performance of a detector since the underlying noise deviates from the assumed model.
• Interference avoidance. Cognitive radio networks should be carefully designed so as to have a smaller degree of interference than the allowable level in licensed user networks.
• Short duration of sensing time (observation time). Sensing time is defined as the time required by a detector to arrive at a certain decision.
• Limited resources: They include the number of users, the control channel bandwidth, the quantization level, etc.
• Various licensed user types: This includes different modulation schemes, data rates, transmission powers, etc.

The general objective of this research project is to investigate and to implement statistical signal processing and communication techniques to tackle some of the above challenges in deriving an efficient spectrum sensing algorithm, which could be divided into two main schemes.

• Fixed sample size. The design of spectrum sensing algorithms is based on the Neyman-Pearson (NP) criterion. This approach leads to the Likelihood Ratio Test (LRT) that fixes the required sample size (sensing time) in order to attain the targeted performance.
• Random sample size. This method is mainly based on a sequential analysis proposed by Abraham Wald in 1943. This scheme results in a random sample size with its average being smaller than the minimum sample size resulting from the NP approach (smaller sensing time).

# Key Research Area

Multi-Scale Modeling and Simulation; Signal Processing

# Supervisors

Prof. Dr.-Ing. Abdelhak M. Zoubir, Signal ProcessingProf. Dr.-Ing. Anja Klein, Communications Engineering

# Contact

Fiky Suratman
M.Sc.

 Address: Dolivorstraße 15 D-64293 Darmstadt Germany Phone: +49 6151 16 - 24401 or 24402 Fax: +49 6151 16 - 24404 Office: S4|10-309 Email: fsuratman (at) gsc.tu...