Interference-aware spectrum sensing mechanisms in cognitive radio networks

Highlights

• Study spectrum sensing mechanisms by the optimization of network parameters and strategies.

• The proposed mechanisms have fewer sensing overhead.

• SUs can access into the spectrum holes efficiently, and transmit with a low power.

Abstract

This paper focuses on the spectrum sensing mechanisms, which could improve network throughput through the sensing strategy optimization and cooperative spectrum sensing methods. In order to guarantee an integrated and effective research, we take the whole channel scenarios into consideration, i.e., Single Secondary user with Single and Multiple Channels (SSSC and SSMC), Multiple Secondary users with Single and Multiple Channels (MSSC and MSMC). Moreover, according to the specific feature of each scenario, different sensing methods are adopted, i.e., optimal sensing period to maximize network throughput for SSSC, a novel sensing method to minimize searching time for SSMC, partial cooperative spectrum sensing mechanism for MSMC, and setting a spectrum pool in the fusion center to record the channel states for MSMC. Simulation results demonstrate that our methods can improve spectrum efficiency, network throughput and channel utilization, especially when the spectrum is utilized inadequately.

Introduction

The rapid development of network services and applications has an increasing demand of wireless spectrum resource, which is actually limited [1], [2], [3]. Conventional access policies only allow licensed users, called Primary Users (PUs), to utilize pieces of licensed spectrum regardless of whether PUs are active or not, which will cause the underutilization of spectrum resource. To address this problem, Cognitive Radio (CR) technology was proposed and advocated [4] for enabling the Secondary Users (SUs) to access and utilize the licensed spectrum. Recently, the US Federal Communications Commission (FCC) has made a decision to permit SUs to access into the licensed spectrum under the condition that no interference is imposed to PUs [5].

Spectrum sensing is a very important function in CR Networks (CRNs) [6], [7]. In order to guarantee the quality of communication, spectrum sensing strategy should be conducted before the spectrum is accessed into SU to detect whether the channel is idle. The reason is that SUs can only access into the spectrum when it is idle, otherwise, the SUs should wait for some time to sense the spectrum again. Generally speaking, spectrum sensing strategy includes the following three methods:

• Transmitter-based sensing method, where the SUs analyze the channel state to judge whether PU has occupied the channel. In order to detect the signals with weak strength, high detection sensitivity is needed, which requires high accuracy of radio frequency front end and analog-to-digital converter [8]. The complexity of this method is low, and prior knowledge of PU is not necessary. The received signals are sampled by time window after fast Fourier transform, and the signal energy can be calculated by a window function to compare with the pre-defined threshold [9]. The main disadvantage of this method is difficult to decide the judging threshold, i.e., the pre-defined threshold, since it is largely affected by the background noise.

If SUs have obtained the information that some signals to be transmitted by PU, matching filter will be adopted to sense whether PU is idle. Through the comparison of the input and output of the matching filter, this method can sense whether PU exists or not [10]. The main advantage of matching filter is the objective sensing probability can be achieved promptly. However, the received information of PU (e.g., bandwidth and modulation type) by matching filter is required to be demodulated. If multiple types of PUs coexist in the network, the complexity is high when sensing the network, since the prior information and phase synchronization are required.

• Interference temperature-based sensing method is proposed by FCC to increase the utilization of wireless spectrum. When PU and SU share the same spectrum to fulfill their communication, the transmission of PU should be satisfied previously. After that, network nodes measure the maximum interference that PU can sustain, and forecast the interference brought by SU to decide whether SU can join into network. The interference power brought by SU to PU can be measured by interference temperature.

The implementation of interference temperature can be conducted by the following three steps [11]: (1) Measure the interference temperature of PU, by which SU can calculate the acceptance level of interference noise, and estimate the interference to PU when accessing into the receiver; (2) set threshold for interference temperature to guarantee the network can work regularly after SU has joined; (3) control the accumulated interference brought by SU to guarantee the interference temperature of PU does not exceed the predefined threshold.

In general, there are three methods to measure the interference temperature [12], [13]: (1) Self-detection method, which senses the transmission power and the interference temperature of SU, then calculates the corresponding interference brought to PU. The SU will terminate the communication if the calculated interference value exceeds the threshold. However, this method will cause the hidden and exposed terminal problems since SU is responsible for the interference control among users; (2) indirect detection method, which senses the interference level continually with fixed monitors. This method converts the interference level to interference temperature and deliveries this level to the transmitter of SU, by which SU can judge whether the frequency is occupied. To guarantee the effectiveness of the sensing value, strong relationship is required between the sensed and actual interference. However, this condition is difficult to be satisfied since the discreteness and randomness in wireless environment will result in the uncertainty of the detected interference level; (3) direct detection method, which senses the interference temperature of PU, and reports this value to each SU. Although this method can overcome the drawback of indirect detection method, it is unrealistic for each PU to have the ability to sense the interference temperature.

• Since the received Signal-to-Noise Ratio (SNR) value is low when the SUs are shielded, they cannot detect the existence of PU reliably. Therefore, SUs may mistakenly consider the channel is idle and access into the channel for packet transmission, which will interfere with PU. In order to solve the problem of hidden terminal, multiple SUs can conduct spectrum sensing in a cooperative method. It has been demonstrated in [14], [15], [16], [17] that the cooperative method can largely increase detection probability in fading channel. The cooperative spectrum sensing method can be briefly divided into the following four steps: (1) Each SU conducts local spectrum detection independently, and makes binary decisions; (2) all SUs deliver their binary decisions to the receiver, which can either be an access point of Wireless Local Area Networks (WLANs) or Base Stations (BSs); (3) the receiver integrates the received binary decisions, and judges whether the frequency band of the observed PU exists; (4) the receiver sends the final judgment to each SU. Since the cooperative spectrum sensing method can reduce time consumption and alleviate the effect on multipath and shadow, it is promising to solve the problem of hidden terminal.

Over the last decades, much work focused on improving the spectrum utilization of communication networks [18], [19], [20], [21], [22]. The authors in [19] utilized the average capacity of SU as the performance metric to investigate the random allocation scheme for both general and Rayleigh channel fading models. In [20], the authors studied the opportunistic spectrum sharing problem through the joint optimization of spectrum detection and transmission power control, and adopted a multi-state spectrum sensing model for detection decision. However, previous literatures mainly focused on algorithm design for spectrum sensing, and the spectrum sensing mechanism has not been well studied. Due to the limitation of detection ability, the constraints of hardware cost and power consumption, it is difficult to increase the spectrum utilization by merely adopting spectrum sensing algorithms. Therefore, spectrum sensing mechanisms are adopted to solve this problem. Our objectives are to discover the idle spectrum fleetly and increase network throughput under the condition that the interference of PU does not exceed a certain threshold. In this paper, we focus on the spectrum sensing mechanisms by the sensing parameter optimization, sensing strategy selection and partial cooperative spectrum sensing method to improve the achievable network throughput. Network performance in four scenarios are studied, namely: Single Secondary user and Single Channel (SSSC), Single Secondary user and Multiple Channels (SSMC), Multiple Secondary users and Single Channel (MSSC), and Multiple Secondary users and Multiple Channels (MSMC).

The rest of this paper is organized as follows: Section 2 introduces the spectrum sensing mechanisms for SSSC and SSMC. The spectrum sensing mechanisms for MSSC and MSMC are described in Section 3. Section 4 presents the numerical results, and Section 5 concludes this paper.