# Introduction he radio spectrum is among the most heavily regulated and expensive natural resources around the world. In Europe, the 3G spectrum auction yielded 35 billion dollars in England and 46 billion in Germany. The question is whether spectrum is really this scarce. Although almost all spectrums suitable for wireless communications has been allocated, preliminary studies and general observations indicate that much of the radio spectrum is not in use for a significant amount of time, and at a large number of locations. According to the statistics of the Federal Communications Commission (FCC), temporal and geographical variations in the utilization of the assigned spectrum range from 15% to 85% [1]. The limited available radio spectrum and the inefficiency in spectrum usage necessitate a new communication paradigm to exploit the existing spectrum dynamically. The FCC has realized that overcrowding of unlicensed bands is only going to worsen and is considering opening up licensed bands for opportunistic use by secondary radios/users. One of the most efficient paradigms is the cognitive radio network (CRN), first introduced by J. Mitola [2], which is built upon software-defined radio (SDR) technology. S. Haykin defines a cognitive radio (CR) as "an intelligent wireless communication system that is aware of its environment and uses the methodology of understanding-by-building to learn from the environment and adapt to statistical variations in the input stimuli" [3,4] with the overarching aim of providing reliable communication whenever and wherever needed while efficiently using the resources available to it. A CRN is built on the following principle: a network of secondary users (SU) (users without license) continuously senses the use of a spectrum band by primary users (PU) and opportunistically utilizes the band when PUs are absent. Any SU in a CRN performs two main functions: (1) sensing spectrum usage to identify absence or presence of a PU, and (2) T transmitting at appropriate power if the PU is absent. Thus, the design principle for CRN regards the CR users as visitors in the spectrum they occupy (Fig. 1). The successful operation of these principles relies on the CRN users' ability to be aware of their surroundings, which is accomplished through spectrum sensing solutions. Spectrum sensing (the optimization of sensing parameters in a single spectrum band, spectrum selection and scheduling, and an adaptive and cooperative sensing method) enables CR users to adapt to the radio environment by determining currently unused spectrum portions, so-called spectrum holes or spectrum white spaces, without causing interference to the primary network. Generally, spectrum sensing techniques can be classified into four groups: (1) primary transmitter detection (matched filter detection, energy detection, and feature detection), (2) cooperative detection, (3) primary receiver detection, and (4) interference temperature management [1]. The SUs periodically sense the spectrum in order to identify and use the spectrum holes. When a SU senses that the PU is accessing the spectrum or determines that the PU is going to access the spectrum, it then vacates the spectrum and moves to the another spectrum or lowers its transmitting power. If there is another SU instead of the PU, then the first SU shares the spectrum with the new SU. Spectrum holes exist both in time and in space. A spectrum hole in time may arise when a PU of the spectrum is idle, i.e., not transmitting. In this case, temporal spectrum hole is the duration for which the primary user-transmitter (PU t ) is in the idle state (Fig. 2 (a)). A spectrum hole in space with respect to a given frequency channel may occur if a given SU is sufficiently far from a PU that is actively transmitting (Fig. 2 (b)). In this case, the SU may transmit up to a certain level, which we called the maximum interference-free transmit power (MIFTP), without causing harmful interference to PUs who are receiving the transmissions. This results a useful spectrum sharing. However, in the typical scenarios of static CRN where SU observe the activity of the PU in a fixed spectrum band and access the entire spectrum band if a spectrum opportunity is detected (based on IEEE 802.11 and IEEE 802.15.3 technologies), a SU transmitter communicates directly with SU receiver only when the PU is idle. A SU has to stop its transmission when no spectrum holes exist (Fig. 2 The SU transmitter communicates directly with the SU receiver, due to the existence of a (a) temporal spectrum hole or a (b) spatial spectrum hole with respect to PU t . # Typical CRN transmission limitation due to lack of spectral holes in shown in (c) To improve the transmission capacity of CRN, here we approach to combine CRN with cooperative communication strategy employing joint spatialtemporal sensing to maximize the transmission capacity of SUs in a CRN. In these scenarios, when PU transmitter is active, SU transmits to another SU receiver via a relay channel. By enabling the use of both the direct and relay channels, the transmission performance of the secondary system (CRN) can be improved significantly. Our proposed approach is depicted in Fig. 3. # SU-Transmitter Actively Transmitting # SU-Receiver Actively Receiving # PU-Transceiver Active # NO Communication # SU-Transmitter Actively Transmitting # SU-Receiver Actively Receiving # PU-Transceiver Active or Idle Cooperative communications is a new paradigm that draws from the ideas of using the broadcast nature of the wireless channel to make communicating nodes help each other, of implementing the communication process in a distribution fashion and of gaining the same advantages as those found in MIMO systems [5]. The end result is a set of new tools that improve communication capacity, speed, and performance; reduce battery consumption and extend network lifetime; increase the throughput and stability region for multiple access schemes; expand the transmission coverage area; and provide cooperation tradeoff beyond source-channel coding for multimedia communications. This idea is true for wide varieties of wireless networks such as mobile ad hoc networks, sensor networks, or cellular networks. In this paper, we consider the decode-and-forward (DF) cooperative strategy focusing on the case of a single-hop relay channel. # Direct # II. # System model a) Transmission frames and PU t behavior Time on the wireless channel is divided into frames consisting of s N symbols. We shall assume perfect symbol level timing synchronization between the nodes of the secondary system. The primary usertransmitter (PU t ) alternates between active and idle states on a per-frame basis according to the active-idle Markov model of Fig. 4. The active and idle durations can be modeled by geometric random variables with parameters q and p , respectively [6]. We focus on scenarios in which p q > ; i.e., on average, PU t is more often in the active state than the idle state (which is more practical to assume). Let us suppose that kp q = , where k is an integer with 1 > k . Hence, if PU t is idle for one time frame on average, it will be active for K time frames, on average. A similar approach can be used when q p > . # b) Cooperative transmission protocol Suppose a secondary user-transmitter (SU t ) desires to transmit s N symbols; i.e., it requires one full frame in which PU t is idle. The cooperative protocol works as follows (Fig. 3): When PU t is active, SU t sends information to secondary user-receiver SU r via the relay nodes SU relay . When PU t is idle, SU t sends information directly to SU r . In order to achieve this, the secondary nodes (SU t , SU relay and SU r ) perform joint spatialtemporal sensing [7]. In particular, all secondary users estimate their MIFTPs based on signal strength measurements, which they exchange with one another. They also decide whether the PU t is active or idle, by transmitting their local decisions to a fusion center, which then makes the final decision. We shall assume that there exists a communications path from SUt to SUr through SUrelay. This path would be established by a routing protocol. A repetition code [8] is used to repeat the transmission of the signal over K 0 and 1 K frames, where K K K = + 1 0 . At each relay, the transmitted signal is amplified and forwarded to the next relay node until it reaches the destination. When PUt is idle, SUt can communicate s N symbols over one time frame directly to SUr. Due to broadcast wireless channel, SUrelay also received a copy of signal from SUt but ignores it. When SUr receives a signal from the relay path, it stores the received signal and waits until it receives the same signal directly from SUt and vice versa. The received signals at SUr are then combined using MRC. Finally, maximum likelihood (ML) detection is used to detect the signals. # c) Cooperative Relaying The received signal of a simple wireless channel model with flat fading and no shadowing is given by [9] ( ) n hs P n s ? h d r d ? y + = + = ? (1) Where 2 ? is the free space signal-power attenuation factor between the source and a reference distance r d , d is the distance between the source and destination, ? is the propagation exponent, ( ) 2 0 h ,? h~C? is a complex Gaussian random variable with variance 2 h ? , ( ) 0 0,N n~C? , and s is the transmitted signal. In Eq. ( 1), ( ) ? ? ? d r d P 2 = denotes the equivalent transmitted power after taking into account the effect of path loss. We also define ( ) ( ) 0 0 2 0 ? ? ? d r d P = , ()1n s P g K i i n s P i g i g y 0 0 1 0 * 1 + ? = = + = ,(2) Where ? = = 0 1 2 ~K i i g g and ? = = 0 1 * ~K i i i n g n , s is the transmitted symbol, 1 2 = s and i g is the channel gain between SU t and SU relay during time frame . The received signal at SU r due to relay SU relay is ( ) ( ) R n s g A h P P K j j n n s P g A j h P j h R y + = ? = + + = 1 0 1 1 0 1 * , (3) Where ? = = 1 1 2 K j j h h , ? = + = ? ? ? ? ? ? 1 1 * 1 2 K j j n j h n P A j h R n , j h is the channel gain between SU relay and SU r during time frame j . Here, A is the amplification factor which is chosen to maintain average constant power output at SU relay , ( ) g N g P A 0 2 0 1 2 + = . The noise variance of R y is 0 0 2 1 2 2 N h N h g P A R + = ? where ? = = 1 1 2 ~K j j h h . The direct transmission (SU t ? SU r ) channel model is T n s t P f T y + = , (4) Where f is the channel gain between SU t and SU r . ( ) ( ) r t R P P h g A N t P f w ? ? ? ? + = + = 2 1 0 2 0 2 ,(5) Where ( ) 0 2 N t P f t = ? and( ) ( ) ( ) 1 1 0 1 0 2 1 0 2 ~+ + = = ? ? ? ? ? ? R P P h g A r ,(6) with ( ) 0 0 0 N P g = ? and ( ) 0 1 1 N P h = ? . We assume that f , i g , j h are known at receiving end. The symbol error probability (SEP) conditioned on the instantaneous SNR is given by ( ) [10] where k is a constant that depends on the type of modulation and w k Q e P ? =( ) ( ) ? ? ? = x t e x Q dt 2 / 2 2 1 ? is the standard Q -function. III. # Performance analysis In this section, we derive a lower bound on the average SEP. We show that the SEP lower bound is minimized when 1 0 K K = for K even and 6), we can upper-bound r ? as follows: 1 1 0 + = K K for K odd. From Eq. (( ) ( ) ( ) h g N P P r 0 2 1 0 2 1 0 1 0 1 0 = ? + ? ? ? ? ? ? ? ? ,(7) Taking expectations on both sides, we have [ ] ( ) [ ] h g E N P P r E 0 2 1 0 ? ? ,(8) Note that ) , ~ CN( j , h i g 1 0 then g 2 and h 2 are independent 2 ? -distributed random variables with 0 2K and 1 2K degrees of freedom, respectively. Applying Jensen's inequality, [ ] [ ] [ ] [ ] 1 0 2 2 2 1 ~K K h E g E h g E h g E = = ? ,(9) From Eq. ( 8) and Eq. ( 9), we have [ ] ( ) 1 0 0 2 1 0 K K N P P r E ? ? ,(10) Assuming M-PSK modulation, the average SEP can be expressed as follows [11]: ( ) ( ) ( )d? 1 0 1 ? ? ? ? ? ? ? ? ? ? = r M M M t M SEP ,(11) Where 1 0 K K e r E e r e E r M ? ? ? ?? ? ? ? ? ? = ? = ? ,(12) Where ( ) 0 0 2 1 0 ? ? N P P ? ? . The lower bound for the SEP is then obtained by substituting the right hand side of Eq. ( 12) into Eq. ( 11). If K is even, i.e., m K 2 = where m is an integer, m K K K K 2 1 0 1 0 2 = + ? , with equality holding when m K K = = 1 0 . In this case, Eq. ( 12) implies that the choice of 1 0 K K = maximizes the performance of our proposed scheme. If K is odd, i.e., 1 2 + = m K , information-theoretic results in [12] suggest the choice 1 0 K K > , in order to maximize the SNR at the first hop. Let n K K + = 1 0 , where 1 ? n is an integer. We have -n m K 1 2 1 2 + = , m m -n m m K K 4 2 4 2 1 4 2 4 1 0 4 + ? + + =(13) The equality in (13) holds for 1 1 0 + = K K , which also suggests that choosing 1 1 0 + = K K minimizes ( ) ? ? ? r M and hence the SEP. Our analysis is confirmed by the simulation results presented in Section 4. IV. # Numerical results In this section, we present the simulated result of our proposed scheme in CRN. In the performance curves, the 95% confidence intervals are omitted for clarity. BPSK modulation is used and ( ) 1 , 0 , , CN j h i g f The frame length is 100 symbols. MRC and detection are used at the receiver. In the first phase, we have assumed that the three channels have equal average SNRs, i.e. the distances between source, relay and destination are assumed to be equal. So it is logical to think the same MIFTPs for this case. , the transmission strategy follows the scheme described in Section 2, which we call (strategy 1). The traditional cooperative communication scheme, which we call strategy 2, transmits m times on the path SU t ? SU relay ? SU r over m branches. Fig. 5 shows the SEP performance of our scheme with one relay and In practice, the situation in the CRN becomes so much complex that it leads us to think different distances among the SU t , SU relay and SU r (instead of assuming equal distances among them having the same average SNRs) which results different fading characteristics of these three channels. Hence these three channels will have different average SNRs. The simulated result of this condition is presented in Fig. 7 when the MIFTPs and the average SNRs ( ) . This also confirms the results in [12] in which the maximum transmission capacity is achieved when the relay is situated slightly near the source terminal or in the middle between the source and the destination. # Conclusion We proposed a scheme works with the cooperative communication strategy in cognitive radio networks so that the secondary users can get all time transmission connectivity by using both spatial spectrum white space sensing and temporal spectrum white space sensing. This results maximum possible spectrum utilization. The proposed scheme improves the transmission capacity of static cognitive radio network to a great extent. 1![Figure 1 : CRN architecture. The CR users coexist with the primary users. The CR users can use both unlicensed bands and licensed bands that are free from primary users' activities](image-2.png "Figure 1 :") 2![Figure 2 : Typical spectrum sharing scenarios in CNR:](image-3.png "Figure 2 :") ![Enhancing Transmission Capacity of a Cognitive Radio Network by Joint Spatial-Temporal Sensing with Cooperative Communication Strategy](image-4.png "E") 3![Figure 3 : Joint spatial-temporal spectrum sensing in a typical CRN for improving the transmission limitation due to lack of spectral holes by utilizing cooperative communication](image-5.png "Figure 3 :E") ![inequality and Eq. (10), we have Enhancing Transmission Capacity of a Cognitive Radio Network by Joint Spatial-Temporal Sensing with Cooperative Communication Strategy ( ) [ ] [ ]](image-6.png "E") 5![Figure 5 : Decode-and-Forward (DF) with one relay and K = 4](image-7.png "Figure 5 :") 4![Fig. 5 shows the SEP performance of our scheme with one relay and 4 K = in comparison to pure temporal sensing, pure spatial sensing, and traditional cooperative communications with 1 1 0 = = K K . The pure temporal sensing scheme corresponds to direct communication from SU t to SU r , whereas the pure spatial sensing scheme corresponds to communication over the relay path. The simulation results confirm our analytical conclusion that the best performance is achieved with our proposed scheme with 2 1 0 = = K K . Similar performance trends can be seen in Fig. 6, for which 5 = K , 3 0 = K and 2 1 = K .](image-8.png "4 K") 6![Figure 6 : DF with one relay and K = 5](image-9.png "Figure 6 :") ![We found that a higher SNR at the link from SU t to SU relay results the best performance. The valuable insight from this is that we should choose the relay node that maximizes the average SNR ( )](image-10.png "") 7![Figure 7 : DF with one relay, asymmetric SNR, and K = 4](image-11.png "Figure 7 :") Enhancing Transmission Capacity of a Cognitive Radio Network by Joint Spatial-Temporal Sensing withCooperative Communication Strategyactive and t ? denotes the maximum transmit power ofSU t . When? << 0?t0 K time frames [8]013as2Year22D D D D ) E(q1-qactiveidle1-pp( d powers from SU process ) t t d r t P ? ? ? 2 =1 d as the equivalent transmitted 2 1 ? ? ? r d P = and© 2013 Global Journals Inc. (US)t to SU relay , from SU relay to SU r , and from SU t to SU r , respectively. Here, 0 d , 1 d and t d denote the distances between the node pairs (SU t , SU relay ), (SU relay , SU r ), and (SU t , SU r ), respectively. Also 0 ? and 1 ? denote the MIFTPs of SU t and SU relay , respectively when PU t is Figure 4 : 2-state Markov chain model for PU t active/idle R y and T y . The noise variables R n and T n havedifferent powers becauseR n includes a noisecontribution at the relay. For this reason, noisenormalization is necessary for MRC of T y and R y as in[10]. The resulting SNR is © 2013 Global Journals Inc. (US) * YXiao FHu Cognitive Radio Networks New York CRC-Taylor & Francis Group 2009 * Making Software Radios More Personal JMitola GQMaguire CognitiveRadio IEEE Personal Commun 6 4 1999 * JGProakis Wiley Encyclopedia of Telecommunications Hoboken, New Jersey John Wiley & Sons, Inc 2003 * Cognitive Radio: Brain-empowered Wireless Communications SHaykin IEEE J. Select. Areas Commun 23 2005 * KJ RLiu AKSadek WSu AKwasinski Cooperative Communications and Networking 2008 Cambridge University Press * AMotamedi ABahai MAC protocol design for spectrum agile wireless networks: Stochastic control approach -Proc. IEEE DySPAN'07 2007 * TDo BLMark Joint spatial-temporal spectrum sensing for cognitive radio networks-Proc, Conf. on Information Sciences and Systems (CISS'09) Baltimore, MD Mar. 2009 * DTse PViswanath Fundamentals of Wireless Communication Cambridge Cambridge University Press 2005 * Multihop diversity in wireless relaying channels JBoyer DDFalconer HYanikomeroglu IEEE Trans. Commun 52 10 2004 * Symbol error probabilities for general cooperative links ARibeiro XCai GBGiannakis IEEE Trans. Wireless Commun 4 2005 * MKSimon M.-SAlouini Digital Communication over Fading Channels 2004 John Wiley & Sons, Inc 2nd Edition * Capacity theorems for wireless relay channels GKramer MGastpar PGupta 41st Allerton Conf. on Commun 2003