^{1}

^{1}

^{2}

^{1}

^{2}

We propose a more practical spectrum sensing optimization problem in cognitive radio networks (CRN), by considering the data traffic of second user (SU). Compared with most existing work, we do not assume that SU always has packets to transmit; instead, we use the actual data transmitted per second rather than the channel capacity as the achievable throughput, to reformulate the Sensing-Throughput Tradeoff problem. We mathematically analyze the problem of optimal sensing time to maximize the achievable throughput, based on the data traffic of SU. Our model is more general because the traditional Sensing-Throughput Tradeoff model can be seen as a special case of our model. We also prove that the throughput is a concave function of sensing time and there is only one optimal sensing time value which is determined by the data traffic. Simulation results show that the proposed approach outperforms existing methods.

Cognitive radio (CR) is a promising technology to exploit the unused spectrum bands licensed to the primary user (PU) [

There is a fundamental problem called Sensing-Throughput Tradeoff which has been studied in depth [

Some important results have been given in related references. References [

Though this problem has been extensively studied with significant results, to the best of our knowledge, existing researches all make an impractical assumption that SU has countless data packets to transmit all the time. Then the optimal sensing time problem is equivalent to the following problem: how to maximize the idle channel capacity in a frame time.

However, this assumption does not always hold because the amount of data packets depends on the traffic flow in practical systems. Actually, it is certainly not SU has countless packets to transmit all the time in practical scenarios. In methodology, the above assumption simplifies the sensing time optimization problem, because the data traffic would have a great influence on the optimal sensing time. It should be noted that the ultimate purpose of all the study on spectrum sensing and access is to complete the data traffic of SU.

In this paper, we discard the impractical assumption that SU has data packets to transmit all the time. We define the SU throughput as the actually transmitted data in the actual time spent, rather than the channel capacity in the frame time and reformulate the Sensing-Throughput Tradeoff problem based on the data traffic of SU. We propose a more general model and prove that the throughput is a concave function of sensing time and there is only one optimal sensing time value which is determined by the data traffic.

The rest of this paper is organized as follows. The system model is described in Section

We consider a slotted frame structure like in [

Frame structure.

We, specially, discard the impractical assumption that SU has data packets to transmit all the time and focus on the actual SU throughput defined by transmitted data in the actual time spent, rather than the channel capacity obtained. In Figure

Obviously, our model is more practical because it is certainly not SU has countless packets to transmit all the time in practical scenarios. Actually, our model is more general compared to the existing one. The traditional model in [

In existing works, under the assumption that all the SUs have countless data packets to transmit all the time, the throughput would be computed by

The above definition of throughput is actually the available channel resource, not the exact throughput obtained by SU.

In order to study the actual optimal sensing order in practical systems, as in Figure

This value is the actual throughput for the data traffic

We adopt energy detection in the sensing duration as in [

Following the analysis in [

We focus on the throughput of SU when PU is inactive, as in [

We focus on the first situation, that is,

There is a tradeoff between sensing and transmission. When

The objective is to find the optimal sensing time

In problem (

There exists only one optimal sensing time

From (

According to (

When

Applying

In addition, according to (

Since

This result means that condition

Combining our research and study in [

If

Else,

In all, there is only one optimal sensing time

In addition, the optimal sensing time in our model is related to the traffic

In this section, we present the simulation results to evaluate the performance of the proposed method. Without loss of generality, similar with [

The bandwidth of PU channel is 6 MHz, the SNR of SU’s signal without PU’s single is 0 dB, and the sampling frequency is the same as the bandwidth. The probability of channel idle is

We assume that the model of data traffic of SU is constant bit rate (CBR) traffic [

Figures

The optimal sensing time,

The optimal sensing time,

Figures

The optimal sensing time,

The optimal sensing time,

Figures

The traffic versus throughput,

The traffic versus throughput,

The traffic versus throughput,

The traffic versus throughput,

The results show that the real throughput for traffic obtained by our method is obviously better than the method in [

When the frame time and SNR value are different, the optimal sensing time and the optimal throughput change. Importantly, the performance of the proposed method is always better than the method in [

In addition, when the traffic is heavy, the data queue may be almost full; the throughput becomes almost the same, which verifies that the model in [

In this paper, we investigate the Sensing-Throughput Tradeoff problem in CRN, considering the effect of traffic of SU. We reformulated this problem by removing the assumption that the SUs always have countless data packets to transmit. The proposed model is more suitable for practical systems and can be regarded as a more general model compared with the existing work, because the traditional Sensing-Throughput Tradeoff model can be seen as a special case of our model. We also prove that the throughput is a concave function of sensing time when the probability of false alarm is less than 0.5, and there is only one optimal sensing time value which is determined by the data traffic. Simulation results show that the proposed method achieves better performance than the existing approaches.

The authors declare that there is no conflict of interests regarding the publication of this paper.

This work was supported by the National Natural Science Foundation of China (Grant nos. 61172062 and 61301160), by Jiangsu Province Natural Science Foundation (Grant no. BK2011116), and in part by the National Basic Research Program of China (Grant no. 2009CB320400). This work was also supported by the National Science Foundation of China under Grant no. 61401508.