Effects of Malicious Users on the Energy Efficiency of Cognitive Radio Networks

Effects of Malicious Users on the Energy Efficiency of Cognitive Radio Networks Efe F. Orumwense 1, Thomas J. Afullo 2, Viranjay M. Srivastava 3 School of Electrical, Electronic and Computer Engineering, University of KwaZulu-Natal, Durban, 4041, South Africa. 1 efe.orumwense@gmail.com 2 Afullot@ukzn.ac.za 3 viranjay@ieee.org Abstract Multiple Cognitive Radio Users (CRUs) perform local spectrum sensing independently and makes a binary decision about the spectrum occupancy. The binary decision is forwarded to a Fusion Center (FC) for fusion which leads to more accurate detection of licensed signals in the network. Some malicious users in the network may affect the decisions of other legitimate CRUs by sending wrong spectrum occupancy information to the FC so as to use the spectral band for their own selfish gain, a term known as Primary User Emulation Attack (PUEA). In this paper, the effect of malicious users on the energy efficiency of cognitive radio networks is examined. A secured energy detection cooperative spectrum sensing technique is proposed and analysed to help maximize energy efficiency and reduce the effects of these attacks on the network. Simulation results show that there is a decrease in the energy efficiency of the network when the malicious users increases, and also that our proposed secured Cooperative Spectrum Sensing (CSS) technique in the OR fusion rule provides better energy efficiency. Keywords Cognitive Radio Networks, Energy Efficiency, Malicious Users, Primary User Emulation Attacks, Fusion Rule. I. INTRODUCTION With the recent advances in wireless communication technology, Cognitive Radio (CR) has gradually paved its way into modern day technology becoming more popular by the day and evolving to become an alluring and attractive solution to spectral congestion and shortage problems [1]. In a Cognitive Radio Network (CRN), a Cognitive Radio User (CRU) or unlicensed user can opportunistically use an unused portion of spectrum belonging to a licensed user without a license. In order to avoid interference with the licensed user during this process, the CRU should have a prior knowledge about the status of the spectrum, either it is being used or vacant before using it. This prior knowledge about the status of the spectrum is gained through a process called spectrum sensing [2]. In a spectrum sensing process, each unlicensed user is equipped with cognitive radio to detect a targeted licensed spectrum and logically decide if the spectrum is free or vacant. In the vein to enhancing the performance of the process, multiple CRU can effectively cooperate with each other to conduct spectrum sensing, a term in literature commonly known as Cooperative Spectrum Sensing (CSS) [3] [4]. Cooperative spectrum sensing allows each cognitive radio to perform local spectrum sensing independently and then makes a binary decision and forwards this decision to a Fusion Center (FC). The FC gathers the local sensing information and makes a final decision about the availability of the spectral band. It has been seen that with the introduction of cooperative spectrum sensing in cognitive radio networks, the effects of multi-path fading and shadowing experienced by CRUs are mitigated and the performance of spectrum sensing has greatly improved with a more accurate detection of licensed user signals [5]. CSS, however, is vulnerable to some misbehaving CRUs which disrupt the network and the spectrum sensing etiquette and the obtainable overall performance. The misbehaviour is caused by reporting false spectrum occupancy information in order to influence the final decision made by the FC. A malicious CRU usually sends information that the spectrum is used to the FC which helps to identify the spectrum as used in taking the final decision. The resultant effect of this is that other CRUs will identify the malicious CRU as a licensed user thereby vacating the occupied spectrum band for the malicious CRU believing it is a licensed user. This gives the malicious user an unrivalled access to the spectrum [6]. An attack in the cooperative spectrum sensing process of a cognitive radio network in which malicious users pretends to be a licensed user by sending false spectrum occupancy information in order to gain unrivalled access to a vacant spectrum is called Primary User Emulation Attacks (PUEA) [6]. One of the possible approaches in preventing PUEA in the network is to build a secure link between CRUs and FC in order to be sure that only authenticated spectrum sensing occupancy results obtained from a trusted CRU is accepted by the FC in making its final decisions. In order to build a secured link, authentication, integrity and accurate spectrum sensing mechanism, are taken into consideration. The spectrum sensing mechanism used must be able to accurately detect vacant spectrum bands. The FC receiving the spectrum occupancy information should be able to attest that information is coming from a legitimate CRU (authentication) and also that the information was not modified or changed in transit (integrity). Due to the large overhead required in most spectrum sensing mechanisms, previous works on this aspect are mainly based on licensed user detection techniques and intrusion detection techniques with cryptographic mechanisms. In [7], an authentication of the licensed user s signal using cryptographic and wireless link signatures via a helper node which is usually placed in close proximity to the primary user is used. Chen et al [8], have investigated a cooperative spectrum sensing scenario in the presence of a Page 431 Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2015

PUEA using optimal weights to maximize the detection probability. Althunibat et al [9] have also proposed an approach that tackles security in the cooperative spectrum sensing process of a cognitive radio with respect to energy efficiency by using a low-overhead symmetric cryptographic mechanism which produces a message authentication code that authenticates spectrum occupancy information sent by CRUs. In [10], a collaborative spectrum sensing protocol was proposed to improve the energy efficiency of the network by reducing the number of sensing reports from the secondary users to the fusion center. The relative trustworthiness of a user in a cognitive radio network is also evaluated in [11] where the authors proposed a method in verifying if the source of spectrum occupancy information is from a legitimate user. In this paper, we have analyzed the effects of malicious CRUs on the detection accuracy of cooperative spectrum sensing process and the achievable energy efficiency by the legitimate CRU. We present that these effects depend mainly on the number of malicious users in the network and the type of Fusion Rule (FR) employed by the spectrum sensing process in making the final decision on the spectrum utilization. The contributions of this paper extend to propose an improved energy detection mechanism that reduces the effects of malicious users on the energy efficiency of cognitive radio network using an improved energy detection based cooperative spectrum sensing mechanism that maximizes the energy efficiency of the network. The organisation of the paper is structured as follows, Section II presents a system model of a cognitive radio network and the resulting effect of malicious users on the detection performance is also formulated. The proposed energy detection based cooperative spectrum sensing mechanism is presented in section III. Simulation results obtained are presented and discussed in Section IV. Finally, sections V conclude the work and recommend future works. II. SYSTEM MODEL Consider a system with number of CRUs trying to access a licensed spectrum band in a cognitive radio network environment. The probability that the spectrum is not being used by a licensed user is denoted by. To avoid interference, each CRU senses a specific spectrum at a specific time and takes a local binary decision * + about the availability of the spectrum. If the decision is taken then the CRU decides that the spectrum is being used. If otherwise, then the CRU decides that the spectrum is vacant. After a local binary decision have been taken by all the CRUs and sent to the FC, the FC receives these decisions and fuses them by using specific fusion rules (FR) to make a final decision. There are many fusion rules that can be applied at the FR [12]. In this work, we will be focusing on the logic OR rule and the logic AND rule because in these rules, given a targeted probability of detection or a targeted probability of false alarm, each CRU threshold can be derived. In OR rule, the FC will declare the spectrum busy when at least one of the CRU detects a licensed user signal, otherwise the spectrum band is regarded as vacant. In the AND rule, the spectrum band is declared busy by the FC only when all the CRU detect the licensed user signal, otherwise the band is regarded as vacant. However, there are some CRUs that do not normally follow the spectrum sensing etiquette and therefore downgrade the overall performance of the considered CRN. These users are referred to as malicious CRUs. What the malicious user does for its own self gain is to always report a spectrum sensing decision of 1 so that the decision of the FC can be influenced. This however increases the probability of the FC in taking a final decision of 1 and therefore, none of the legitimate CRUs will use the spectrum. The malicious user then uses the spectrum for its transmission. We consider the number of malicious user to be so that the total number of CRUs in the network, both legitimate and malicious becomes. In a CSS environment, the probability of detection and the probability of false alarm are used to measure the local performance of the CRUs in the network. Probability of detection is the probability of identifying a used channel as used while the probability of false alarm is the probability of identifying a vacant channel as used. The overall performance of CSS is also measured in a similar manner by the FC using the local spectrum decisions sent by the CRUs. The probability of detection and the probability of false alarm employing the OR fusion rule in terms of M and N can be expressed as { (1) { (2) The probability of detection and the probability of false alarm employing the AND fusion rule in terms of M and N can also be expressed as { (3) { (4) where is a predefined threshold on the number of CRUs who detect a user on the spectrum. is the probability of detection for each individual CRU. In equation (1) and (2), when, the final decision will be 1 since the FC will always receive a number of 1 s larger than or equal to K reported by the malicious users. But when, the lower bound summation will be decreased by. and are the probability of detection and false alarm respectively in the local spectrum sensing process of any of Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2015 Page 432

the th CRU in the cognitive radio network. This can be expressed as and ( ), (5) ( ), (6) where indicates that th CRU has decided that a licensed signal is present, and indicates the presence and absence of a licensed user signal respectively. The effect of malicious users on the overall performance of the network can be seen in and. An increase in creates a more efficient network while a decrease in degrades the efficient usage of the unused spectrum. In formulating the effects of malicious users on the network s resources, we define Energy efficiency metric as the ratio of the total successfully transmitted bits to the total number of energy consumed by the legitimate CRUs. Data can only be successfully transmitted if the spectrum is vacant and no false alarm is reported, which results to non-zero efficiency. The average energy efficiency can be expressed as where the factor represents the probability of no false alarm case, represents the data rate in, is the transmitted time, is the energy consumed by all the legitimate CRUs during sensing and is the energy transmitted by the scheduled CRU. From equation (7), an increase in the false alarm probability will lead to a decrease in the amount of successfully transmitted data which on the other hand lowers the energy efficiency. III. SECURE ENERGY DETECTION BASED COOPERATIVE SPECTRUM SENSING In order to reduce the effects of malicious users on a CRN, it is imperative to build a secured CSS which will help to increase the energy efficiency of the system. This section presents an energy efficient cooperative spectrum sensing method which also helps to increase the detection of licensed signals in the targeted spectrum band. Since a malicious user is present in the system to always report the presence of a licensed signal where there is actually none, then fake signals will be sent by a PUEA and received by other legitimate CRUs under only. So in the event of an attacker, only the probability of false alarm will be affected. So involving the presence or absence of an attacker and respectively in equation (6), we then have (7) ( ) ( ), (8) where and are conditional probabilities regarding the presence and absence of fake PUEA attacker signals which are related to the attacker strategy. Considering and as constant values, for simplicity of notations, we can denote is as, (9) and, (10) where denotes the presence of an attacker. We can therefore rewrite equation (8) as ( ) ( ) (11) A. Energy Detection Based Spectrum Sensing Technique Energy detection sensing technique is very popular in cooperative spectrum sensing due to its simplicity and no requirement on a prior knowledge of the licensed user signal [13]. A local spectrum sensing is performed by the all the CRUs, both legitimate and malicious. It is assumed that every CRUs adopts the energy detection technique in which G samples of the energy are summed up during a detection interval,. (12) is compared to a threshold which every CRUs decides locally about the presence and absence of a licensed user signal. The probability of detection and the probability of false alarm for an th CRU in energy detection can be written as: ( ), (13) ( ), (14) where is the threshold used in energy detector of the th CRU. Based on equation (12), in energy detection is sum of squared of the CRU received signal. is Gaussian random variable with zero mean and variance under, * +. So will be compliant with the central Chisquare distribution with 2 degrees of freedom and parameter. { * + * + * + (15) where is the possible outcome of the presence of a licensed user signal, is the presence of a PUEA signal and is the presence of none of the signals. In determining the analyzed cooperative spectrum sensing method used, we employ Neyman-Pearson criterion [14] to determine the probability of detection using energy detection based cooperative spectrum sensing. Neyman-Pearson technique provides a threshold for detection subject to a constant probability of false alarm. Based on equation (8), we need the values of ( ) and ( ), which can be written in energy detection as ( ) ( ) (16) ( ) (17) Page 433 Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2015

So we can now rewrite in equation (5) energy detection based cooperative spectrum sensing as, (18) where and are Gamma function and upper incomplete Gamma function [15], respectively. Equation (11) can also be written as Parameter TABLE I PARAMETERS USED FOR SIMULATION Value 0.5 0.8 0.2 0.3s. (19) In Neyman-Pearson criterion, it is shown that for a given probability of false alarm, the optimal threshold which maximizes the probability of detection can be obtained if the given probability of false alarm is the actual considered probability of false alarm. The probability of detection and the probability of false alarm based on our method for both the OR and AND fusion rules are reformulated respectively, as follows (20) (21) Energy Efficiency. (Bit/Joule) 3000 2500 2000 1500 1000 Proposed Secured Algorithm (AND Rule) Normal CSS Algorithm Proposed Secured Algorithm (OR Rule) (22) (23) where and can easily be obtained by replacing by in equation (1) (4) respectively. By the same way, the achievable energy efficiency in our proposed secure CSS can be updated as follows 500 1 2 3 4 5 6 7 Number of malicious users(m) Fig. 1. The achievable energy efficiency against the number of malicious users in a cognitive radio networkfor the normal and proposed secured CSS algorithm. (24) 3500 3000 where, is the energy consumed by one CRU during the local spectrum sensing and is the energy required to report its one bit spectrum occupancy information to the FC. IV. SIMULATION RESULTS We have implemented the simulations on a cognitive radio network consisting of 10 legitimate users. The simulation parameters used as regards the energy efficiency and network specifications are given in Table 1. Energy Efficiency. (Bit/Joule) 2500 2000 1500 1000 500 OR Fusion Rule AND Fusion Rule 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Probability of false alarm (Pfa) Fig. 2. The achievable energy efficiency against the probability of false alarm in a cognitive network for the proposed secured CSS algorithm in the OR and AND Fusion rules. Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2015 Page 434

Fig. 1 shows the effects of malicious users on the energy efficiency of a cognitive radio network. The attainable energy efficiency is plotted against the number of malicious users in the system. It is evident that as the malicious users continue to increase in the network, the energy efficiency decreases. Observe that there is higher energy efficiency with the proposed energy detection based cooperative spectrum sensing technique in both the AND and OR fusion rules than the normal cooperative spectrum sensing technique. That is to say that there is an improvement in the energy efficiency achieved by the proposed secure algorithm in both rules over the normal cooperative spectrum sensing protocol. The results in Fig. 2 shows the performance of the OR and AND fusion rules of the proposed secured energy detection based cooperative spectrum sensing technique. The energy efficiency achieved is plotted against the probability of false alarm. As the probability of false alarm increases, the energy efficiency of the network also decreases. The increase in the probability of false alarm is caused by malicious CRUs taking wrong spectrum occupancy decisions thereby affecting the number of legitimate users in the network. The result shows that there is a better performance of the OR fusion rule over the AND fusion rule. V. CONCLUSIONS AND FUTURE WORKS Primary User Emulation Attacks (PUEA) populates a cognitive radio network with malicious users which affect the energy efficiency of the network. In this paper, the effect of these malicious users on the network is studied. A secured energy detection based cooperative spectrum sensing technique is proposed to boost the security of the spectrum sensing process of CRUs in the network where the final decision about a spectrum occupancy is taken by a Fusion Center (FC) employing Fusion Rules (FR). The simulation results show that malicious users has a great impact on the energy efficiency of the network and our proposed secure CSS algorithm has a significant improvement on the attainable energy efficiency when compared to the normal CSS algorithm. Also from the two fusion rules employed by the FC, the OR fusion rule yields more energy efficiency than the AND fusion rule. In future work, the effects of imperfect reporting and sensing channels on the energy efficiency of the network will be investigated. [5] A. Ghasemi and E. S. Sousa, Collaborative spectrum sensing for opportunistic access in fading environments, in Proceedings IEEE Symp. New Frontiers in Dynamic Spectrum Access Networks (DySPAN 05), Baltimore, USA, pp. 131-136. November,, 2005. [6] E. Orumwense, O. Oyerinde, and S. Mneney, Impact of primary user emulation attacks on cognitive radio networks, International Journal on Communications Antenna and Propagation, Vol. 4, pp. 19 26. April, 2014. [7] Y. Liu, P. Ning, H. Dai, Authenticating Primary Users Signals in Cognitive Radio 1209 networks via integrated cryptographic and wireless link signatures, in Proceedings Of IEEE Symposium on Security and Privacy, California, USA, pp. 286 301, May, 2010. [8] C. Chen, H. Cheng, Y-D. Yao, Cooperative spectrum sensing in cognitive radio networks in the presence of the primary user emulation attack, IEEE Transactions on Wireless Communications. vol. 10. pp. 2135-2141. 2011. [9] S. Althunibat, V. Sucasas, H. Marques and J. Rodriquez On the trade off between security and energy efficiency in cooperative spectrum sensing for cognitive radio. IEEE Communication Letters vol. 17, no. 8, pp. 1564-1567. July 2013. [10] S. Mousavifar and C. Leung, Energy efficient collaborative spectrum sensing based on trust management in cognitive radio networks IEEE Transactions on Wireless Communications. vol. 14, pp. 1927-1939. 2015. [11] E. Orumwense, O. Oyerinde and S. Mneney, Improving trustworthiness amongst nodes in cognitive radio networks Proceedings of the Southern Africa Telecommunication Networks and Applications Conference (SATNAC), Eastern Cape, South Africa. pp. 401-406. August 2014. [12] S. Kyperountas, N. Correal, and Q. Shi A Comparison of Fusion Rules for Cooperative Spectrum Sensing in Fading Channels. EMS Research, Motorola. [13] I. Akyildiz, B. Lo, and R. Balakrishnan, Cooperative spectrum sensing in cognitive radio networks: a survey Journal on Physical Communications. vol. 11. pp. 40-62. 2011. [14] Z. Yuan, D. Niyato, H. Li, and Z. Han, Defense against primary user emulation attacks using belief propagation of location information in cognitive radio networks, in proceedings of the IEEE Wireless Communications and Networking Conference (WCNC). Hong Kong,. pp. 599-604. March, 2011. [15] Gradshteyn and I. Ryzhik, Table of integrals, series and products, 6 th edition. New York. Academic Press, 2000. Efe Francis Orumwense received his B.Sc (Hons) degree from the School of Engineering, University of Benin, Benin City, Nigeria in 2009. He also received a Master s degree from the School of Electrical Electronic and Computer Engineering, University of KwaZulu-Natal, Durban, South Africa in 2014. He is currently working towards a PhD from the same institution. His research interest includes Orthogonal frequency division multiplexing systems, cognitive radio technology, wireless network security and energy efficiency systems. REFERENCES [1] J. Mitola and G. Q. Maguire, Cognitive radio: Making software radios more personal IEEE Personal Communications, vol. 6, no 4, pp. 13-18, August 1999. [2] S. M. Mishra, A. Sahai, and R. Brodersen, Cooperative sensing among cognitive radios, in Proceedings IEEE International Conference in Communications. Istanbul, Turkey. June 2006. vol. 4, pp. 1658-1663. [3] K. B. Letaief, and W. Zhang, Cooperative communications for cognitive radio networks, in Proceedings of the IEEE Journal, vol. 97. 2009. pp. 878-893. [4] Y. Zou, Y.D Yao and B. Zheng, Cooperative relay techniques for cognitive radio systems: Spectrum sensing and secondary user transmissions IEEE Communication Magazines, vol. 50, no. 4, pp. 98-103, April, 2012. Page 435 Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2015