Keywords: Green Networks; Cognitive Radio; Cooperative Beamforming; ABC algorithm; Benders ... demarcated as an intelligent wireless communication system which is cognizant of the environment and its changes ..... [3] Simon,Haykin.
Available online at www.sciencedirect.com
ScienceDirect Procedia Technology 24 (2016) 888 – 895
International Conference on Emerging Trends in Engineering, Science and Technology (ICETEST - 2015)
Combined Optimization of Clustering using Artificial Bee Colony Algorithm and Cooperative Beamforming in Green Cognitive Radio Networks with Relays Ashok S Kumara , Sudha T.b a M-Tech b
Student, Electronics and Communication Engineering ,NSS College of Engineering Palakkad 678008, Kerala, India Professor, Electronics and Communication Engineering, NSS College of Engineering, Palakkad 678008, Kerala, India
Abstract In this paper, our objective is to reduce the interference in the direction of Primary Users (PUs) and to decrease the power cost of Secondary Users (SUs) in the Cognitive underlay system. Artificial Bee Colony (ABC) algorithm, one of the most recently introduced meta-heuristic optimization algorithms is used here as the clustering algorithm. Here relay stations are introduced in cooperative beamforming technique, which are equipped with directional antennas that can efficiently improve the spectral efficiency and can reduce the interference to primary users in a cognitive underlay systems. In the proposed research, flexible cooperation is considered as a general case while full cooperation and inter-cell cooperation are considered as special cases. We jointly optimize the beamforming and the clustering to reduce the overall power consumption of the secondary users and to satisfy the primary users interference limits. Inorder to find optimal solution, an iterative algorithm called generalized Benders decomposition method is applied. We have investigated the benefits of ABC algorithm and showed the advantages of the relays in cognitive wireless networks. c 2016 2016The TheAuthors. Authors.Published Published Elsevier Ltd. © byby Elsevier Ltd. This is an open access article under the CC BY-NC-ND license Peer-review under responsibility of the organizing committee of ICETEST - 2015. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of ICETEST – 2015 Keywords: Green Networks; Cognitive Radio; Cooperative Beamforming; ABC algorithm; Benders decomposition algorithm; Indirect relay mode ;
1. Main Text Cognitive Radio (CR) is observed as an innovative approach for improving the utilization of limited natural resource: the radio electromagnetic spectrum. CR is a revolutionary evolution of Software Defined Radio (SDR). It is demarcated as an intelligent wireless communication system which is cognizant of the environment and its changes and can adjust its transmission/receiver parameters accordingly. CR has two primary intentions in mind: highly re-
剷 Corresponding author.
Tel.: +919497360455 ; E-mail address: a s h o k s k 1 8 1 2 @ g m a i l . c o m
2212-0173 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of ICETEST – 2015 doi:10.1016/j.protcy.2016.05.157
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
889
liable communication whenever and wherever needed and resourceful utilization of the radio spectrum. CR has the ability to exploit the unused spectrum (e.g. TV White Spaces) in an intelligent way while not interfering with other users. It advances the spectrum utilization by exploiting these unused spectrum called spectrum holes [1]-[4]. Green communication is energy efficient communication [5]. Inorder to achieve green communication, an attempt is made in this work to reduce the interference pollution in the direction of primary users and total power cost of secondary. Spectrum sensing is the capability to measure, sense and be aware of the parameters related to the availability of spectrum, radio channel characteristics, and transmit power, interference and noise. The detection performance is often compromised with shadowing, multipath fading and receiver uncertainty problems. To alleviate the impact of these factors, cooperative spectrum sensing is found to be the best sensing method. An outline of cooperative spectrum sensing has been studied in [6]. In cooperative sensing, different users are in different radio environments, and some of the users are in good position for acquiring each specific spectrum hole [7]. In our approach, we initialized many nodes (secondary users). Nodes are divided into different clusters according to the clustering algorithm and select a CH for each cluster. Some of them can access spectrum hole easily. We used cooperative spectrum sensing as the sensing method to sense spectrum hole. In our endeavor centralized cooperative spectrum sensing method is used, in which there is a Cluster Head (CH) node that collects the information’s from other nodes. Cooperative Beamforming technique has been studied extensively in [8]. This technique can improve the range of the communication process, since the signal beam is concentrated only to the direction of the intended partner so that no energy is wasted to the other direction. So Inorder to reduce the interference to PU, we have espoused this technique in our work. Multi-cell cooperation schemes can be categorized into three types like Flexible cooperation, Inter-cell cooperation, Full cooperation [11]. Here we consider flexible cooperation scheme as general case and inter-cell coordination and full cooperation as special cases. The work in [11] jointly optimize the beamforming and clustering to minimize the overall power consumption while rewarding the secondary users QoS and satisfying the primary users interference temperature limits. This problem is expressed as a mixed-integer nonlinear program and decomposes it into a master problem and a beamforming sub problem. An iterative algorithm that has been derived based on Benders decomposition in [11] is used to find the optimal solution in this work. This algorithm is a way to divide complex mathematical programming problems into two, and thereby disentangle the solution by solving one master problem and one sub problem [12]. Amplify and forward (AF) relaying used in [13] can increase the throughput and progress the sensing performance. In our work we consider the cases for without relay and with relay. In the Indirect (Relay) mode, the relays are equipped with omnidirectional receiving ability and directional transmission capability. So message signal from one CH is collected by relay node and forward to another CH, thereby reduces the interference to primary users and improves spectral efficiency [14]. The work reported in [9],[10] exploited the advantages of using Artificial Bee Colony algorithm (ABC) as compared to other clustering algorithm. ABC is a population based global optimization algorithm which is motivated by the intelligent foraging behavior of honey bees is used here as the clustering algorithm. There are several advantages of ABC algorithm which makes it more attractive than other optimization algorithms like fast convergence, simplicity, less control parameters and robustness. It can also be integrated with other optimization algorithms easily. To the best of our knowledge, ABC algorithm has not been used so far in Multi cell cooperation schemes. We make the following assumptions like perfect channel information and perfect synchronization for the cooperative beamforming. The position of nodes is assumed to be static, that is the position does not change with time. The position of each node is elected randomly following a uniform distribution. The nodes are sufficiently separated so no overlapping of nodes takes place. 2. System Model 2.1. ABC algorithm for clustering The ABC comprises of three groups : employed bee, onlooker bee and scout bee. The bee going to the food source which is visited by itself previously and sharing the collected information with other bees is employed bee. The bee
890
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
waiting on the dance area for making decision to choose a food source is onlooker bee. The bee resounding random search is scout bee. Unemployed bees are scout along with onlooker bee. The unemployed and employed bees search for the rich food sources around their hive. The location of nodes is chosen randomly and this nodes corresponds to food source analogously in the clustering context [9]. Similarly the position of a better solution to the given optimization problem is indicated by the position of a good food source and the fitness cost of the associated solution is represented by the quality of nectar of a food source. With the help of this better solution and fitness cost, cluster groups and cluster heads are selected. ABC algorithm has four phases [9]: Initialization phase Initial food source position (location of nodes) of S Z solutions is generated randomly in this phase, where S Z denotes the size of employed bees or onlooker bees. Each solution ai, where i = 1, 2...S Z is a D-dimensional vector representing the number of optimized parameters. The location of nodes is randomly produced using the following expression: ai, j = amin, j + rand(0, 1)(amax, j amin, j )
(1)
where amax and amin are the upper and lower bound of the parameter ai , respectively and j denotes the dimension. And then evaluate each nectar amount famti . Employed bees phase In this phase, each employed bee ai finds a new food source qi in its neighborhood. The fitness of food sources is also calculated. Onlooker bees phase Employed bees share their information about food sources with onlooker bees waiting in the hive and then onlooker bees probabilistically choose their food sources depending on this information. After that probability of food source is calculated. Then the onlooker bees will carry out random search in their neighbourhood. The new solution produced is compared with the current solution and the better one is memorized by means of a greedy selection mechanism similar to the employed bees. Scout bees phase Employed bees whose solutions can’t be improved through a predetermined number of cycles, called limit, become scouts and their solutions are abandoned. Then, they find a new random food source position using the following expression: ai, j = a j,min + D.(a j,max a j,min (t))
(2)
where D is a random number lies between 0 and 1. These steps are repeated through a predetermined number of cycles called Maximum Cycle Number (MCN) or until a certain termination criterion is satisfied. Steps for ABC Clustering algorithm 1. 2. 3. 4.
Cycle=1 Initialize the food source positions ai , 1, ...S Z Evaluate the nectar amount fitness function f amti of food sources Repeat
891
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
5. Employed Bees Phase: For each employed bee Produce new food source positions qi Calculate the value f amti Apply greedy selection mechanism End 6. Calculate the probability values for the solution 7. Onlooker Bees Phase: For each onlooker bee Choose a food source depending on probability value Produce new food source positions qi Calculate the value famti Apply greedy selection mechanism End 8. Scout Bee Phase: Check if an employed bee becomes scout, if so replace it with a new random source positions 9. Memorize the best solution achieved so far and cycle = cycle + 1 10. Until cycle =Maximum Cycle Number. 2. Cooperative Beamforming in green Cognitive radio networks In this work cognitive underlay system is considered in which the secondary users are represented by and primary users, order to model the system a binary vector is defined:
. Cognitive BSs,
has m antennas each. In
(3) The component
of
determines whether BS,
serves the secondary user, s or not. (4)
In full cooperation each user can be served by many BSs. The special case of full cooperation is (5) in which each user is collectively served by all BSs. The special case of inter-cell coordination corresponds to: (6) which ensures that each user is assigned to only one BS. Assuming full cooperation, Let represents the channel vector from the distributed antennas of all BSs to the receiving antennas of a secondary user, s.
is the channel between BS, and user, s. The channel vector
fading and path loss attenuation. Similarly,
corresponds to both Rayleigh
represents the array beamforming vectors
associated with u s e r , s a n d is the beamforming vector of BS, for user, s. The special case of full cooperation in which each user, s is served by all BSs is considered here, it’s received signal is represented as: (7)
892
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
Where
represents the conjugate transpose,
is a complex zero mean and
user, s that is transmitted by BSs through the channel user,
represents the data symbol of
. The signal to interference plus noise (SINR) ratio of
is given by:
(8)
is the variance. Achievable rate is given by the equation: (9) where W and Γ represents the channel bandwidth and the SINR gap between the rate achieved in practical case and Shannon capacity. The transmit power,
of each BS, b should not exceed a practical power limit: (10)
For flexible cooperation, a constraint is added: ,
(11)
The role of the additional constraints is to make sure that when a specific BS, b does not serve a specific user, s, the should be a zero vector. SINR definition initially assumes full cooperation; after the constraint beamforming vector is added it is also valid for flexible cooperation. In the cooperative cognitive underlay system the secondary BSs shares the same channel with the primary system under the condition that it should not cross the interference temperature limit imposed by each primary user as : (12) is the channel from secondary BS, b to the primary user,
: (13)
where , denote the total power consumption, the transmit power radiated by all BSs, the signal processing power cost, the fixed power cost and the backhaul power cost respectively. Inorder to collectively optimize clustering and beamforming vectors that minimize the overall power consumption of the secondary system and interference power constraints at the primary users, an iterative algorithm based on the generalized Benders decomposition method in [11],[12] is used to find an optimal solution. 3. Problem Formulation The performance of SU in any CR system depends on the activity of PU and availability of spectrum. Here we consider nodes as secondary users. One of the secondary users is selected as the CH according to the selection procedure. Consider the communication between two secondary users via CH. During this time if PU also starts transmitting, then SU has to reduce its power, otherwise it will cause interference to PU. So the probability of signal to reach the destination decreases. In this work, the aim is to reduce the total power cost of SU while limiting the interference in the direction of PU. 4. Indirect Relay mode equipped with Directional antennas The problems encountered above can be overcome by the introduction of relay node in this model. There are two transmission modes: direct mode and indirect (relay) mode. Indirect mode takes place when the direct link from cognitive transmitter to receiver fails and the process of relaying starts. The SINR value of the relay is compared with a predefined threshold and those relays which will have the value of SINR greater than or equal to the threshold value is chosen to participate at the indirect mode. The received signal of the relay node can be expressed as:
893
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
(14) r= 1,...,N, N is the number of relay node. l=1,...L, L is the number of receivers. message signal to
relay from secondary transmitter.
primary transmitter to
relay node.
and
is the time delay in arriving the
indicates the time delay of the message signal received from
are the message signal transmitted by the primary transmitter and
message signal transmitted by the secondary transmitter received at
secondary user. Similarly
represents the
transmission power of the secondary transmitter to the
SU, continuous power transmitted by primary transmitter
represents the optimization parameter,
is the channel response from cognitive transmitter to the
respectively.
relay. is the noise at the relay node. is the channel loss from primary transmitter to the relay node. SINR value of the relay for the indirect mode is given by: (15) =1,...,L. L is the number of receivers, noise variance,
.
is the SINR value of the relay node of the
is the channel loss from primary transmitter to relay node.
secondary transmitter to the
SU.
cluster,
is the
is the transmission power of the
is the channel response from cognitive transmitter to the
SU and
is the
SU. After determining those relays that are involved in indirect channel response from cognitive transmitter to the transmission, the relaying phase starts to meet the QoS requirement of SUs. In the indirect mode, the received signal at the
SU can be expressed as: +
Also
and
are the channel losses between relay and the
respectively. The parameters phase.
and
(16)
SU and between the Primary transmitter and the
SU
indicates transmission power and message signal of the relay in the relaying
is the time delay of the message signal received at the
signal received from primary transmitter to
,
SU.
is the noise at
SU and
indicates the time delay of the message
SU.
3. Simulation results For the numerical simulation, we consider 20 nodes and 6 Cluster heads (CH) under the assumption that the users are non-overlapping and uniformly distributed. The cluster head holds the information like number of nodes, location of the nodes and energy of those nodes that are involved in the specific cluster. For simplicity, we makes the assumption that SINR target are same for all users i.e. ρ s = ρ, s . Fig. 1.(a) and Fig. 1.(b) depict the cluster formation plot using ABC algorithm without relays and with two relays respectively. The CH, which holds the information about the specific group is indicated by smaller triangle and Relay Node (Fig.1.(b)) by bigger triangle (green). BS is denoted by circle, which is located at the center of the upper boundary of the square. The total interference temperature limit is taken as Iv = 3dB above the noise level. In the initialization phase of ABC algorithm, the position of nodes is chosen randomly. Similarly the position of a better solution to the given optimization problem is indicated by the position of a good food source and that solution gives the CH. Two cases are considered here. For the first case, no relay has been used and whenever a specific node has to transmit (Fig. 1.(a)), CH which has the information regarding the intended node will select the simplest path to reach the BS. For the second case, relay mode is considered. Here we consider two relays. One of the relay is selected in between the first two CH that are involved in the process of transmission.
894
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
One of the CH that takes part in the communication process is taken as second relay (Fig. 1.(b)). The relay nodes has directional transmission capability and omni directional receiving capability. The first relay node collects the information from first CH and forwards it to the second CH (Fig. 1.(b)). The second CH forwards to next CH and so on to reach the destination. The second relay is one of the CH that takes place in this communication process. During the transmission process, other clusters which are not involved in the process of transmission are considered as primary users.
Fig. 1. (a) Cluster formation plot with ABC algorithm without relays; (b) Cluster formation plot with ABC algorithm using two relays;
Fig. 2. (a) Total power cost without relay; (b) Total Power cost with two relays;
Fig. 2.(a) and Fig. 2.(b) illustrate the total power cost of secondary users without relays and with relays respectively. The total power cost of Full cooperation scheme is high since it requires overhead power cost for sharing channel information and user data through backhaul network. On the other hand this scheme has the lowest transmit power. The total power cost is calculated as the sum of signal processing power, backhaul power, transmit power, overhead power, and fixed power cost of all the clusters involved in transmission process [10]. Fig. 2.(b) shows the total power cost of secondary users with two relays. The relay has less overall power cost. Since one of the CH acts as the second relay, the fixed power cost, backhaul power etc can be reduced. Thus the total power cost of three schemes can be abridged. The total interference to primary users without relays and with relays has been depicted in Fig. 3.(a) and Fig. 3.(b) In this work, the performance of flexible cooperation, inter-cell cooperation and full cooperation schemes has been studied. As the value of SINR increases, interference to primary user also increases. While considering Full cooperation scheme, the interference to primary user is fewer for values of ρ < 10dB and rises afterwards as compared to flexible cooperation and inter-cell cooperation schemes. The inter-cell cooperation is optimal at smaller SINR value. When the value of ρ > 10dB, it is no longer feasible. Fig. 3.(a) shows the plot without using relay. The relay has directional transmission capability, which can direct the signal to the intended destination, thereby reduces the interference in the direction of primary users. Also the probability of signal to reach the intended destination increases. By using relays, for the three multicellular cooperation schemes, the total interference to primary users can again be reduced.
Ashok S. Kumar and T. Sudha / Procedia Technology 24 (2016) 888 – 895
895
4. Conclusion The objective of this work is to reduce the interference in the direction of Primary Users and to decrease the power cost of Secondary Users in the Cognitive underlay system. Centralized cooperative spectrum sensing is the spectrum
Fig. 3. (a) Total interference to primary users without relay; (b) Total interference to primary users with two relays;
sensing technique used in our work which can improves sensing performance significantly. The clustering algorithm used in our work is ABC algorithm. In the proposed research, flexible cooperation is considered as a general case while full cooperation and inter-cell cooperation are taken as special cases. We jointly optimize the beamforming and the clustering to reduce the overall power consumption of the secondary users and to satisfy the primary users interference temperature limits. Inorder to find optimal solution an iterative algorithm called generalized Benders decomposition method is used. Relay stations are introduced in cooperative beamforming technique. We assert that relay technique can reduce the interference to primary users and improves the spectral efficiency, since it has directional transmission capability. Our results disclosed the advantages of implementing relays for this scheme, for reducing total power cost of secondary users and interference to primary user. References [1] Ian F, Akyildiz, Won-Yeol Lee,Mehmet C. Vuran, Shantidev Mohanty. NeXt Generation/Dynamic Spectrum Access/Cognitive Radio Wireless Networks: A Survey, Computer Networks 2006; vol. 50, no. 13, p. 2127-2159. [2] K. C. Chen, Y. J. Peng, N. Prasad, Y. C. Liang, and S. Sun. Cognitive Radio Network Architecture: Part I - General Structure, ACM International Conference on Ubiquitous Information Management and Communication (ICUIMC) 2008. [3] Simon,Haykin. Cognitive Radio: Brain-Empowered Wireless Communications ,Selected Areas In Communications, IEEE journal on 2005; p. 201-220. [4] Shrikrishan Yadav, Santosh Kumar Singh, Krishna Chandra Roy. A Smart and Secure Wireless Communication System: Cognitive Radio, International Journal of Soft Computing and Engineering 2012 (IJSCE); Vol.2, Issue-1, p. 2231-2307. [5] Ziaul Hasan, Hamidreza Boostanimehr and Vijay K. Bhargava. Green Cellular Networks: A Survey, Some Research Issues and Challenges, IEEE Communication Surveys Tuts 2011; vol.13, p. 524-540. [6] Ian F. Akyildiz, Brandon F. Lo, Ravikumar Balakrishnan. Cooperative spectrum sensing in cognitive radio networks: A survey, Physical Communication 4 2011; p. 40-62. [7] Xiaoming Chen, Hsiao-Hwa Chen, and Weixiao Meng, Cooperative Communications for Cognitive Radio Networks:From Theory to Applications, IEEE Communications surveys and tutorials 2014; vol.16, no.3, p. 1180-1192. [8] Yu Zhang, Anese and Georgios B. Distributed Optimal Beamformers for Cognitive Radios Robust to Channel Uncertainties, IEEE transactions on signal processing 2012; vol.60, issue 12, p. 6495-6508. [9] Twinkle Gupta and Dharmender Kumar. Optimization of Clustering Problem Using Population Based Artificial Bee Colony Algorithm: A Review, International Journal of Advanced Research in Computer Science and Software Engineering 2014; vol.4, p. 491-502. [10] Dervis Karaboga and Celal Ozturk. A novel clustering approach: Artificial Bee Colony (ABC) algorithm, Applied Soft Computing 2011; vol.11, p. 652-657. [11] Rindranirina, Ramamonjison, Alireza Haghnegahdar and Vijay K. Bhargava. Joint Optimization of Clustering and Cooperative Beamforming in Green Cognitive Wireless Networks, IEEE Transaction on wireless communication 2014; vol.13, no.2, p. 982-997. [12] J. Benders. Partitioning procedures for solving mixed-variables programming problems, Computational Management Science 2005; vol. 2, no. 1, p. 3-19, Available: http://dx.doi.org/10.1007/s10287-004-0020-y [13] Narges Naouri, Narges Noori. Directional Relays for Multi-Hop Cooperative Cognitive Radio Networks, Radioengineering 2013, vol.22, no.3, p. 791-799. [14] Yulong Zou, Jia Zhu, Baoyu Zheng, and Yu-Dong Yao. An Adaptive Cooperation Diversity Scheme With Best-Relay Selection in Cognitive Radio Networks, IEEE transactions on signal processing 2010; vol. 58, no. 10, p. 5438-5445.