Survey PaperComparative performance analysis of evolutionary algorithm based parameter optimization in cognitive radio engine: A survey
Introduction
Over the past decades, the demand for electromagnetic spectrum has increased exponentially due to the popularity of wireless devices such as smart phones and mobile devices. Since most of the usable spectrum is already allocated, the demand for more spectrum is a huge challenge for researchers worldwide. Previous studies have shown that 90% of the allocated spectrum is either unused or underutilized [1], [2]. So the solution lies in the efficient usage and higher utilization of the available spectrum. The present spectrum policy uses static spectrum allocation which has resulted in underutilization of the spectrum [1]. Most of the spectrum bands are only used in some areas and for some part of time. Hence researchers have proposed a dynamic spectrum allocation approach which allocates spectrum dynamically depending on the need at a particular time or location. Cognitive radio (CR) uses this technique to provide better spectrum efficiency and utilization.
The CR is a wireless radio that senses its electromagnetic environment and dynamically adapts its operational parameters to achieve the best system performance. The CR was first proposed by Mitola and Maguire [3] as a software radio with radio knowledge based reasoning about radio etiquettes such as RF bands, protocols, and patterns. The important feature of the CR includes sensing, learning and adapting its operating parameters depending on the radio environment, primary and secondary user requirements, availability of spectrum, local radio protocols, etc. Currently most of the licensed bands use a single communication technology. But the CRs use a combination of different technologies to get the best performance at a particular instant of time. This requirement poses a major challenge for the industry and academic researchers. The other major challenges include spectrum sensing, architecture, engine design, cognitive network security, etc. Recently the CRs have attracted the interests of various researchers [4], [5] and led to applications such as wireless innovation alliance and IEEE 802.22 standard in wireless regional area network [6].
This paper focuses on the challenges in designing an efficient CR engine. The CR engine configures the radio system parameters to provide the best performance with respect to a predefined set of objectives and constraints. Since the objectives in a wireless environment are multiple in number, the problem of CR engine design can be formulated as a multiobjective optimization problem. The literature on the design of CR engine can be broadly categorized based on single and multicarrier communication system, and single and multiobjective optimization.
The techniques mentioned in this section describe the CR engine design only for single carrier communication systems. These techniques can be further categorized based on single and multiobjective optimization.
In this case, the multiple objectives are weighted independently to generate a single objective fitness function which is simple and easy to implement from hardware point of view. The weights assigned to the objective functions mostly correspond to three modes of operation: emergency, low power and multimedia. The first CR engine was designed by the researchers at Virginia Tech [7], [8]. They have used genetic algorithm (GA) to adapt the radio parameters of the software defined radio in a time varying radio environment. Rieser [9] has proposed a GA based CR engine in which the objectives are weighted differently depending on the user requirement and condition of the radio link. This weighted sum GA based design considered code rate, operating frequency, modulation type, etc. as the design parameters and frame error rate, transmit power, spectral efficiency, etc. as the design objectives. Hauris [10] has used the GA for changing the radio parameters of CRs in autonomous vehicles which travel through rapidly changing radio environments. Kim et al. [11] have designed a software to model a CR system. They have incorporated spectrum sensing ability and a GA based CR engine into the software. Zhang and Xie [12], [13] have proposed a CR engine model using neural network in which the learning and adaptation of the CR is based on the fixed as well as the variable factors. The sensitivity of different transmission parameters on the performance of a GA based CR engine has been studied in detail by Newman and Evans [14]. An ant colony optimization (ACO) based CR engine design is proposed by Zhao et al. [15]. Huynh and Lee [16], [17] have used a two-dimensional structure for chromosome’s implementation while optimizing the parameters of CR engine.
Park et al. [18] have proposed a goal-Pareto based nondominated sorting genetic algorithm (GBNSGA) for the CR engine design. They have validated the application of the GBNSGA in a code division multiple access (CDMA) 2000 forward link by using a realistic scenario in a Rician channel. A multiobjective immune GA (MIGA) based CR is proposed by Yong et al. [19]. They have used the MIGA for designing a control module based on the IEEE 802.11a physical platform.
The research activities in mobile communication have shown that multicarrier communication systems provide higher data rate and robustness to losses in the wireless channel. Orthogonal frequency division multiplexing (OFDM) is one of the multicarrier systems which provides flexible resource allocation among cognitive users. Any subcarrier in the OFDM can be deactivated easily by feeding zero power to it. This feature makes the OFDM a good candidate for the CRs [20]. A lot of studies have been carried out for resource allocation in OFDM systems [21], [22], [23].
Newman et al. [24] have designed a weighted sum GA based CR engine for OFDM based transceivers. They have derived a set of objective functions for guiding the search direction of the GA. They have also demonstrated the trade off between the convergence time and size of the search space of the GA. The CR engine design based on different variants of the GA is also proposed in literature [14], [25], [26], [27], [28], [29]. Most of the OFDM based CR engine designs do not consider the existence of primary users. But in a real time scenario, the primary and the secondary users exist simultaneously and use different communication access technologies too. This leads to mutual interference which ultimately degrades the performance of both the users [20]. In literature, mutual interference in the OFDM based CR is considered only in single objective optimization problem such as capacity maximization [30], [31], [32] while ignoring secondary user’s performance.
The recent studies have shown that the new evolutionary algorithms (EAs) perform better than the GA in terms of quality of solution, convergence time and computational complexity. Zhao et al. [33] have proposed a CR engine design based on binary particle swarm optimization (PSO). They have shown that the PSO based CR engine design performs better than a GA based counterpart in terms of fitness value, convergence speed and stability. Afterwards, a PSO based CR engine in real number space [34] has been proposed to improve the performance in decision making. El-Khamy et al. [35] have shown that a hybrid of binary-coded PSO and GA (HBPGA) for optimizing the radio parameters of a CR performs better than the conventional GA and PSO. Waheed and Cai [36] have used the binary ACO for adapting the parameters of a CR in multicarrier environment. They have shown that the binary ACO based CR engine design provides a better solution than a GA based counterpart. Waheed and Cai [37] have compared the performances of the binary PSO, binary ACO and GA in the CR engine framework and found the binary PSO to be a better candidate among the three algorithms. Huang et al. [38] have proposed a CR learning engine based on support vector machine and demonstrated its performance on 802.11a protocol platform. Chen and Wen [39] have introduced cross entropy method for optimizing the parameters in the CR engine design. They have demonstrated that their proposed method performs better than the conventional PSO based CR engine design. Pei-Pei et al. [40] have used a weighted sum based GA for resource allocation in an OFDM based CR by considering bit error rate, throughput and transmission power as objective functions. However they have not considered the spectral interference introduced by the secondary user into the primary user’s band. Di et al. [41] have suggested the use of immune theory as a priori knowledge to guide the EAs in optimizing the parameters of a CR engine. Zhang et al. [42] have introduced a binary quantum-behaved PSO for optimizing the parameters of a CR engine. Artificial bee colony (ABC) algorithm based parameter adaptation for a CR engine has been proposed by Pradhan [43].
El-Saleh et al. [44] have proposed an adaptive CR decision engine based on a multiobjective hybrid GA to determine the optimal radio parameters for a single carrier system. Chen et al. [45] have proposed several schemes such as population adaptation, variable quantization, variable adaptation, multiobjective GAs, etc. to enhance the performance of a GA based CR engine. Tosh et al. [46] have suggested the use of nondominated sorting based genetic algorithm to determine the necessary transmission parameters of a CR for a multicarrier system in different scenarios. For designing a CR engine, Pradhan and Panda [47] have shown the performances of four state of the art multiobjective evolutionary algorithms (MOEAs), i.e. nondominated sorting genetic algorithm, multiobjective particle swarm optimization, multiobjective bacterial foraging optimization and multiobjective cat swarm optimization. However the complexity involved in multiobjective optimization and hardware constraints have restricted the usage of the MOEAs in real time applications. In addition, the high speed, small size, compatibility, etc. makes the design of a CR engine a big challenge for researchers worldwide.
The present study provides a generalized approach to solve the CR engine design problem using six popular EAs. The quantitative analysis of the simulation results are carried out using different performance metrics. The two types of interference commonly encountered in a wireless environment have been considered. The spectral interference introduced by the cognitive user into the primary user’s band and vice versa are investigated. The possible real time scenarios in a time varying wireless environment are also taken into account during the investigation. The performances of different EAs in single and multicarrier communication systems are compared in each scenario to provide a global view of the CR engine design problem. The qualitative and quantitative analysis of the performances of these algorithms provides an in-depth understanding of the complexities involved in the design problem.
Section snippets
Cognitive radio system model and parameters
The CR model assumes that the primary and secondary users coexist in nearby spectrum bands. Fig. 1 depicts a scenario where the central frequency band B is occupied by the primary user and the sidebands are occupied by the cognitive users. In this study, it is assumed that only one primary user is using the central band B and the sidebands are used by a single cognitive user.
For multicarrier based CR, the OFDM is used as the modulation technique for the cognitive user. As shown in Fig. 2, the
Problem formulation
Based on all the discussion mentioned in previous sections, the CR engine design problem for single carrier communication system can be formulated assubjected to the constraints defined by (11), (12).
Similarly the CR engine design problem for multicarrier communication system can be formulated assubjected to the constraints discussed in Section 2.4.
The weights and
Brief overview of evolutionary algorithms
In the past two decades, population based optimization algorithms have gained popularity in solving complex optimization problems. These algorithms, called bio-inspired techniques, include GA [61], [62], [63], PSO [64], [65], differential evolution (DE) [66], bacterial foraging optimization (BFO) [67], ABC algorithm [68], cat swarm optimization (CSO) [69], artificial immune system (AIS) [70], etc. which also fall under the category of computational intelligence [71] or evolutionary computing
Performance metrics
With the existence of different EAs, it is necessary to quantify the performance of each algorithm. The important goal in an optimization algorithm is to discover the optimal fitness in the least possible number of generations, computation time and fitness function evaluations. Although these are mostly related to each other, but in certain cases these goals are somewhat conflicting in nature. Hence, a single metric cannot decide the performance of an algorithm in an absolute sense. Based on
Simulation result and analysis
The CR models shown in Fig. 1, Fig. 2 are considered in this investigation. The bandwidth of primary user as well as secondary user is 5 MHz. The intermediate frequency of the primary user is 650 MHz. A Rayleigh fading channel with additive Gaussian noise is considered whose mean value is 1 and the PSD is W/Hz. The maximum transmissible power of the cognitive user is 5 W and the maximum interference power tolerable by the primary user is 10 mW. The modulation schemes considered in this study
Conclusion
In this paper, a CR engine is designed based on the recently developed EAs. A weighted sum approach is used for optimizing multiple objectives in a wireless environment. The evolutionary algorithms used in this study are GA, PSO, DE, BFO, ABC, and CSO. The spectral interference between the primary user and the cognitive user has been considered to imitate the practical scenario in wireless communication. Due to limitations on the size and performance of a CR, constraints have been put on
Role of the funding source
One of the authors, P.M. Pradhan, acknowledges the generous funding received from the Ministry of Human Resource Development, Government of India for carrying out this work.
P.M. Pradhan has received the PhD degree from the Indian Institute of Technology Bhubaneswar in 2012. He has received the B.E. degree in electronics and telecommunication engineering from the University College of Engineering, Burla in 2006 and the M.Tech degree in Telematics and Signal Processing from the National Institute of Technology, Rourkela in 2009. His research interests include sensor network, cognitive radio, evolutionary computing and signal processing.
References (119)
- et al.
Pareto optimization of cognitive radio parameters using multiobjective evolutionary algorithms and fuzzy decision making
Swarm Evolution. Comput.
(2012) - FCC, Spectrum policy task force report, in: Federal Communications Commission (FCC’02), Washington, DC, USA, pp....
- M.H. Islam, C.L. Koh, S.W. Oh, X. Qing, Y.Y. Lai, C. Wang, Y.-C. Liang, B.E. Toh, F. Chin, G.L. Tan, W. Toh, Spectrum...
- et al.
Cognitive radio: making software radios more personal
IEEE Pers. Commun.
(1999) Cognitive radio: brain-empowered wireless communications
IEEE J. Select. Areas Commun.
(2005)- et al.
Achievable rates in cognitive radio channels
IEEE Trans. Inform. Theory
(2006) - 802.22 Working Group, IEEE 802.22 D1: Draft Standard for Wireless Regional Area Networks, Technical Report, IEEE,...
- T. Rondeau, B. Le, C. Rieser, C. Bostian, Cognitive radios with genetic algorithms: intelligent control of software...
- D. Maldonado, B. Le, A. Hugine, T.W. Rondeau, C.W. Bostian, Cognitive radio applications to dynamic spectrum...
- C.J. Rieser, Biologically Inspired Cognitive Radio Engine Model Utilizing Distributed Genetic Algorithms for Secure and...
Cognitive radio engine design based on ant colony optimization
Wirel. Pers. Commun.
Multiuser OFDM with adaptive subcarrier, bit, and power allocation
IEEE J. Select. Areas Commun.
Transmit power adaptation for multiuser OFDM systems
IEEE J. Select. Areas Commun.
Cognitive engine implementation for wireless multicarrier transceivers
Wirel. Commun. Mob. Comput.
Parameters optimization of cognitive radio based on dna genetic algorithm
Fuzzy Inform. Eng.
Implementation of a genetic algorithm-based decision making framework for opportunistic radio
IET Commun.
Cognitive radio adaptation using particle swarm optimization
Wirel. Commun. Mob. Comput.
A novel cognitive radio adaptation for wireless multicarrier systems
IEEE Commun. Lett.
Multi-objective resources allocation for OFDM-based cognitive radio systems
Inform. Technol. J.
Digital Communications
Spectrum pooling: an innovative strategy for the enhancement of spectrum efficiency
IEEE Commun. Magaz.
Cited by (40)
A Reinforcement Learning based evolutionary multi-objective optimization algorithm for spectrum allocation in Cognitive Radio networks
2020, Physical CommunicationCitation Excerpt :To obtain optimal solutions, some researchers transform the MOO problem into an SOO problem [19]. The parameter optimization problem in the CR engine is formulated as a MOO problem, and the authors proposed the WSM to address this problem [20]. In [21], the authors formulate joint multiple relay assignment and source/relay power allocation problem as MOO problem for green cooperative CR networks and adopted a normalized WSM to combine two conflicting objectives to obtain a single objective function.
A survey on secure communication techniques for 5G wireless heterogeneous networks
2020, Information FusionCitation Excerpt :The work provides an overview of different attacks that target the media access. It has been found that most of the schemes have a tendency failure to resist gateway node bypassing attack, node capture attack and user impersonation attack [240,273]. Ik Rae Jeong et al., Han-Yu Lin et al., Yvo Desmedt et al. and Fouad Khelifi (2018) revisited the public-key encryption with keyword search scheme against existential forgery under adaptive chosen-message attacks.
Multi-objective optimization for spectrum sharing in cognitive radio networks: A review
2017, Pervasive and Mobile ComputingCitation Excerpt :Parameter adaptation problem in CRNs is developed as an MOO problem in [51], and an MOEA is suggested in order to address the CR parameter adaptation problem. Pradhan et al. [48] provided a survey on CR parameter adaptation optimization and developed a fitness function related to CR parameter adaption problem. This fitness function is then solved using multiple EAs and the performance of these algorithms is analyzed.
Reinforcement Learning-Based Intelligent Decision-Making for Communication Parameters
2022, KSII Transactions on Internet and Information SystemsA Novel Metaheuristic Optimization for Throughput Maximization in Energy Harvesting Cognitive Radio Network
2022, Elektronika ir ElektrotechnikaOptimization of cognitive radio system using self-learning salp swarm algorithm
2022, Computers, Materials and Continua
P.M. Pradhan has received the PhD degree from the Indian Institute of Technology Bhubaneswar in 2012. He has received the B.E. degree in electronics and telecommunication engineering from the University College of Engineering, Burla in 2006 and the M.Tech degree in Telematics and Signal Processing from the National Institute of Technology, Rourkela in 2009. His research interests include sensor network, cognitive radio, evolutionary computing and signal processing.
G. Panda is currently a Professor of Electrical Sciences at the Indian Institute of Technology Bhubaneswar. He has received the PhD degree from the Indian Institute of Technology, Kharagpur in 1982 and carried out post doctoral work at the University of Edinburgh, UK during 1984–1986. His research interests include digital and adaptive signal processing, intelligent instrumentation, sensor network, soft and evolutionary computing, data clustering, radar signal processing and acoustic noise cancelation.