Article Preview
Top1. Introduction
Both emerging wireless technologies and the static spectrum allocation policy are reasons for the shortage of frequency spectrum. In other words, every old or new operator in the communications market wants to increase its data rate and both need additional spectrum frequency allocation which is often difficult to achieve. Additionally, spectrum allocation policy gives a licensed user (i.e., a primary user) an exclusive access to a specific spectrum and this spectrum cannot be exploited by an unlicensed user (i.e., a secondary user) even if the spectrum is not actually being used by the primary user. So the spectrum should be dynamic instead of static (Akyildiz, Lee, Vuran, & Mohanty, 2006). Now cognitive radio has been proposed in literature to tackle this issue. The term “cognitive” was first defined by Mitola in 1999 (Haykin, 2005; Mitola & Maquire, 1999). It means an intelligent radio that can interact with the external environment and make intelligent behavior – e.g., adjust its transmission parameters such as carrier frequency, bandwidth, power modulation, etc., to tackle changes in the wireless channels. The implementation of cognitive radio through the Software Defined Radio (SDR) concept provides a configurable platform.
Measurements and statistics from the U. S. Federal Communications Commission (2002) show that the exploitation of frequency spectrum is not being fully utilized and only ranges from 15% to 85%. Motivated by this, cognitive radio is a practical solution to spectrum shortage. There exist many tasks for a cognitive radio and among them is spectrum sensing which is the tool that scans the spectrum frequency to find frequency space that is not used by the primary user. After finding this space the secondary user starts its transmission then stops once the primary user starts its transmission.
Many techniques for spectrum sensing have been mentioned in the literature - included are the energy detector (ED), matched filter the (MF) and the cyclostationary detector (Yucek & Arslan, 2009; Bhargavi & Murthy, 2010). The ED is considered the most widely used detector because it is easy to implement in practice. However, it is sensitive to uncertainty in the noise level, and the results in Yucek and Arslan (2009). showed that the ED cannot satisfy a target probability of detection when the received signal to noise ratio is below the SNR wall. The matched filter can be employed when the secondary knows some information about the primary user. For example, a digital TV signal (ATSC), there is a 511-symbol long PN-sequence which can be used. Also, there exists a DC offset of 1.25 is added to the baseband to create a small pilot signal for the carrier recovery. In addition, OFDM system also use preambles for packet acquisition. All these can be used to assist the determination of primary user.
Not a great deal have been done regarding the effect of time offset and carrier frequency offset on the detection performance of spectrum sensing (Notice that we do not take into account the I-Q imbalance because it does not have any effect on the performance of spectrum sensing) . For example, in Zahedi-Ghasabeh, Tarighat, and Daneshrad (2012) the authors studied spectrum sensing in the presence of carrier frequency offset, I-Q imbalance and sampling clock frequency offset for OFDM systems for cyclostationary detector. Other works (Cabric, Tkachenko, & Brodersen, 2006) have studied the performance of the MF in the presence of CFO when the primary user uses a single sine wave pilot. In Cabric, Tkachenko, and Brodersen (2006) the problem of CFO has been addressed by processing coherent segments of the received signal block by block. However, Cabric, Tkachenko, and Brodersen (2006) did not determine how many blocks should be used, where every CFO might require an optimal number of blocks. Moreover, the actual range of CFO where MF outperforms ED has not yet been determined. Also, they did not study the impact of time offset on the performance of the MF.