Adaptive Hybrid Channel Assignment in Wireless Mobile Network via Genetic Algorithm

Adaptive Hybrid Channel Assignment in Wireless Mobile Network via Genetic Algorithm Y.S. Chia Z.W. Siew A. Kiring S.S. Yang K.T.K. Teo Modelling, Simulation and Computing Laboratory School of Engineering and Information Technology Universiti Malaysia Sabah Kota Kinabalu, Malaysia msclab@ums.edu.my ktkteo@ieee.org Abstract In wireless mobile network, the challenge is to assign appropriate frequency spectrum channels to requested calls while maintaining a desirable level of electromagnetic compatibility (EMC) constraint. An effective channel assignment technique is important to improve the system capacity and to reduce the effect of the interference. Most of the existing channel assignment approaches are based on deterministic methods. In this paper, an adaptive hybrid channel assignment (HCA) technique based on genetic algorithm (GA) is proposed. The proposed GA based optimization in HCA scheme is capable to adapt the population size to the number of eligible channels for a particular cell upon new call arrivals in order to achieve faster convergence speed. Besides, the proposed approach can handle both the reassignment of existing calls as well as the allocation of channel to a new call in an adaptive process to maximize the utility of the limited frequency spectrum. The simulation for both uniform and nonuniform traffic distributions on a 49- cells network model show that the average new incoming call blocking or call dropping probability for the proposed hybrid channel optimization method is lower than the deterministic HCA methods. Keywords - evolutionary optimization; genetic algorithm; hybrid channel assignment; wireless mobile network I. INTRODUCTION The extraordinary development and rapid growth of cellular radio network results in the expansion of wireless mobile communications. The rapid growing in the number of users requires a major effort to enhance the performance of wireless communication system. Due to the current development in the communication system is extremely limited by the capacity constraints of the available frequency spectrum, proper utilization of channel allocation techniques which are capable of ensuring efficient channel allocation is essential in solving the channel assignment problem. The channel assignment problem is classified as two types, one of the earlier aims is to assign the required number of channels to each cell in such a way that interference has been precluded and it aims to use the frequency spectrum more efficiently. This problem is called CAP1 and it is NPhard. Another type of channel assignment problem is minimizing the interference while satisfying the demand for channel upon call requests, where this type of problem is called CAP2. CAP2 assumes that interference-free channel assignments do not exist for a given range of frequency spectrums [1]. In general, the channel assignment approaches can be classified into fixed and dynamic. In fixed channel assignment (FCA), the set of channels are permanently allocated to each cell in advance according to predetermined traffic demand. On the other hand, dynamic channel assignment (DCA) refers to the set of available channels assigned dynamically to each cell upon request. The FCA system does not adapt to the change of traffic loads although it is simpler. Therefore DCA surpass FCA in dealing with changing traffic demands, however it has the drawback of consuming more computational effort under heavy traffic load [2]. In order to overcome the drawbacks of the FCA and DCA, a few combinations of the above methods are introduced, for instance the hybrid channel assignment (HCA) and the borrowing channel assignment (BCA). HCA combines the features of both the FCA and DCA schemes. In HCA, the set of available channels is divided onto two subsets, where one set of channels is allocated as the FCA scheme, while the other set is allocated as the DCA scheme. In BCA scheme, the channel assignment is initially fixed as FCA. However the cell will attempt to borrow channel from its neighboring cells when the corresponding cell s channels are fully occupied upon incoming call request to prevent call blocking [3]. Most of the existing channel allocation methods are based on the deterministic methods. These methods require a set of known input parameters and rules to predict the channel assignment results. However, the channel assignment becomes difficult to be solved by the deterministic methods due to its complexity and computational time issues [3]. Technique of frequency reuse has been proposed to maximize the frequency spectrum capacity in cellular network. Frequency reuse concept comprises of using the same frequency channel simultaneously with other cells subject to the base transceiver station (BTS) distance. However this technique would lead to EMC interferences. Hence an efficient frequency reuse pattern is necessary to 978-1-4577-2152-6/11/$26.00 c 2011 IEEE 511

minimize the interferences. There are numbers of heuristics approaches being suggested to overcome the channel assignment problems based on fixed reuse distance concept such as neural networks (NNs) in [1], simulated annealing (SA) in [4], Tabu search (TS) in [5] and genetic algorithm (GA). The evolutionary algorithm (EA) approaches such as GA outperforms other methods such as NNs, SA and TS, in terms of the ability to explore information over search spaces [6]. This type of algorithm can be used to solve complicated optimization task, such as optimal-local, multi-constrained and NP-complete problems [7]. GA originates from the principal of natural selection and survival of the fittest. The algorithm is capable to obtain solutions in highly-nonlinear problems, which are characterized by multimodal solution space [8]. GA can be defined as highly parallel mathematics algorithm by [9] which transforming the population, each with an associated fitness value, into a new generation using operations based on the theories of evolution. Several GA-based approaches have been used to solve the channel assignment problem. For instance, [10] defined an asexual crossover and a special mutation to solve the channel assignment problem. However such crossover will easily destroy the structure of current solution and thus, causing the algorithm difficult to converge. In addition, [11] represented the channel assignment solution as a string of channel numbers which are grouped in such a way that each cell has a specified number of channels rather than using binary string. The results satisfied the traffic requirement. The evolution is then advanced with a partially matched crossover operator (PMX) and basic mutation. In this paper, HCA optimization algorithm based on GA will be presented to solve the channel assignment problem which is categorized as CAP1. The population size of this algorithm is designed to adapt to the number of eligible channels upon new call request for a particular cell, instead of maintaining a fixed population size throughout the simulation. This would ensure that a more reasonable convergence speed can be achieved compared to the fixed population size. II. OVERVIEW OF CHANNEL ASSIGNMENT PROBLEM A. Channel Assignment Constraints The frequency reuse concept in a channel would cause EMC constraint interferences with other channels, which may degrade the quality of the service. There are three types of interference namely co-channel constraint (CCC), adjacent channel constraint (ACC), and co-site channel constraint (CSC). CCC is due to the allocation of the same channel to certain pair of the cells within the BTS distance or reuse distance simultaneously. ACC is caused by the allocation of the adjacent channels to certain pairs of the cells simultaneously and CSC is due to the separation is less than some minimum spectral distance when channels are allocated in the same cell. These EMC constraints are known as hard constraints. There are soft constraints to help in reducing the call blocking probabilities besides the hard constraints. They are the resonance condition, packing condition, and the limitation of reallocation. The resonance condition maximizes the use of channels within the same reuse scheme by allowing the same channels to be assigned to cells that belong to the same reuse scheme. This would greatly reduce the call blocking probabilities. On the other hand, the packing condition allows the use of minimum number of channels each time upon new call requests. Hence it permits the repeated selection of the channels used in other cells as long as the CCC interference is maintained. In DCA, the reallocation process upon a new call arrival will reduce the call blocking probability, but it consumes more time and computation effort. Therefore the limitation of reallocation limits this process being applied only to the cells which involved in new call arrival. This could reduce the situation of excessive reassignment in a cell by attempting to assign the channels which are already assigned before. B. Frequency Reuse Scheme The reuse of channels would cause CCC interference depends on the reuse distance. The channels assigned in different cells need to be separated by a reuse distance which is sufficient enough to reduce the CCC interference to a tolerable level. This ensures that each channel can be reused many times without affected by the CCC interference. The reuse distance is the minimum distance required between the centers of two cells which are using the same channel to maintain the desired signal quality. The distance between the centers of two adjacent cells is considered as a unit distance. The cells with center-to-center distance equals to or multiple of reuse distance belong to the same reuse scheme. Cells may use the same channels within the same reuse scheme. The total number of channel sets that can be formed from the whole frequency spectrum is determined by the number of cells per reuse scheme. Longer reuse distance causes smaller CCC interference level. However, this results in smaller reuse efficiency. Thus the reuse patterns need to be designed by maintaining both the CCC interference level and the reuse efficiency. In this proposed scheme, the co-channel cells are located with a reuse distance of 3 units. This divides the network topology model of 49-cells into seven different reuse schemes. The co-channel cell matrix for reuse scheme is shown in Table I. According to Table I, the co-channel cell matrix consists of a 7 7 matrix with rows represent the y coordinate of the cells and columns represent the x coordinate of the cells. Each cell in the same reuse scheme can be determined with Manhattan distance, where horizontal path length is 2, and vertical path length is 1, means x coordinate moves two units distance and y coordinate moves one unit distance to obtain the required three unit of the reuse distance. In Table I, the two cells belong to the same reuse scheme if the i th row and 512 2011 11th International Conference on Hybrid Intelligent Systems (HIS)

column of the co-channel matrix contains the same number for the two cells. C. Cellular Traffic Model Assumption The proposed cellular traffic model is simulated based on blocked-calls-cleared principle. It means if the entire set of channels in the cellular network is in use in cell involved in new call request and its neighborhood within reuse distance, the call is blocked and dropped with no queuing of the blocked calls. There are 70 channels available in this model to be allocated for incoming calls. The cellular topological model consists of 49 hexagonal cells to form a parallelogram structure, with equal number of cells along both axes. The simulation call traffic distribution can be either uniform or nonuniform distribution. Uniform cellular traffic distribution indicates that every cell has the same traffic load or demand. On the other hand, nonuniform cellular traffic distribution indicates that there is different traffic load in each cell. The nonuniform traffic patterns implemented in this model is shown in Table II. This pattern is used as the initial call rate for simulation. Each of the value represents the average call arrival rate per minute for the corresponding cell. In this case, the average call holding time is 180 seconds. j th III. PROBLEM REPRESENTATION IN HCA SCHEME The channel assignment problem comprises of the assignment of the required channel number to each cell where the interference is avoided. The frequency spectrum is used efficiently with possible reassignment of channel to the ongoing calls in the cell. Assume that a new call arrives in cell k with t 1 existing calls. Then a potential solution vector, V k represents the possible allocation of channels to ongoing calls and the new call at cell k. A chromosome in the genetic representation is expressed as the solution vector of length t, where each gene is a channel number being assigned to a call in cell k. The advantage of this representation is that it consumes shorter computational time to manipulate the vector due to the length of this solution vector is short. In HCA scheme, the whole set of available channels in the system is divided into two sets, which are the fixed set and the dynamic set. When a new call arrives in a cell, the scheme first attempts to serve the call from the fixed set of channels. When all the channels in the fixed set are in use, the system would serve the call with the DCA scheme which is optimized by GA. The simulation of the call arrival event for HCA scheme is shown in Fig. 1. IV. GENETIC REPRESENTATION GA is a useful approach in searching for an optimum solution in the channel assignment problem. It is different Call arrival in cell k y- coordinate TABLE I. CO-CHANNEL CELL MATRIX x-coordinate 1 2 3 4 5 6 7 1 1 5 6 3 2 4 7 Yes Free channel in fixed set of cell k? No 2 4 7 1 5 6 3 2 3 3 2 4 7 1 5 6 4 5 6 3 2 4 7 1 5 7 1 5 6 3 4 4 6 2 4 7 1 5 6 3 7 6 3 2 4 7 1 5 FCA Update FCA allocation matrix No Call blocking Free channel in dynamic set of cell k? Yes TABLE II. y- coordinate NONUNIFORM TRAFFIC DISTRIBUTION (SIMULATION CALLS/MINUTE) x-coordinate 1 2 3 4 5 6 7 1 60 20 15 30 15 60 30 2 60 30 15 30 20 20 60 3 15 30 20 60 60 30 20 Yes Allocate the only channel and update the DCA allocation matrix Is only one channel free? No 4 60 15 20 30 20 30 60 5 20 60 15 60 20 30 20 6 30 20 20 60 30 30 60 Call arrival process completed DCA optimized with GA 7 60 60 15 60 15 20 30 Figure 1. Simulation of call arrival event in HCA scheme optimized with GA strategy. 2011 11th International Conference on Hybrid Intelligent Systems (HIS) 513

from deterministic methods because GA uses randomization. The generic GA which consists of initial population, selection, crossover and mutation, is modified to be implemented with the DCA optimization scheme. A. Initial Population The generation of the initial population for possible channel allocation solutions, V k is the initial stage of the GA algorithm. Upon a new call requests in cell k, a set of eligible channels E(k) is determined in order to assign a possible channel to the new call. In this case E( k) = S ( O( k) P( k)), where S is the entire set of available channels, O(k) is the set of channels allocated to the existing calls in cell k, and P(k) is the set of channels used in the neighboring cells which less than the reuse distance with cell k. All the information related to channel usage can be obtained from the channels allocation matrix A. The initial population consists of the solution vectors with length equals to the magnitude of vector E(k), where each solution contains a unique integer chosen from E(k). Then the remaining (t 1) integers in all the solution vectors are the channels allocated to the ongoing calls in cell k. B. Fitness Function After the generation of the population, a quality measure is necessary to decide the fitness value of each individual among the population, which is called fitness function. As mentioned before, there are soft constraints such as packing condition, the resonance condition and the limitation of reassignment which further reduce the call dropping probabilities besides the hard constraints. These soft constraints are modeled as the fitness function as shown in (1). t F = C. reuse( i, k) Ai, V k, j j = 1i= 1i k k t C k Ai, V k k j j= i= i k dis i k A,, 1 1 (, ) j= 1 k 1 t. (1) where k defines the cell coordinates when a call arrives; t k defines the total number of channels allocated to cell k; C defines the total number of cells in the network model; V k defines the solution vector for cell k with dimension t k ; V k,j defines the j th element of vector V k ; A i,vk,j defines the element at i th row and V k,j -th column of the channels allocation matrix A; dis(i,k) defines the distance between cells i and k; reuse(i,k) defines a function that returns a value of 0 if the cells i and k belong to the same reuse scheme, otherwise return as 1. In (1), the first term represents the resonance condition, where the fitness value increases if the j th element of vector V k, j V k is used in cell i where cells i and k does not belong to the same reuse scheme. The second term represents the packing condition where the fitness value decreases if the j th element of vector V k is used in cell i, and cells i and k are free from CCC interference. The fitness value decreases with the distance between cells i and k. On the other hand, the last term represents the limiting reassignment condition, where the fitness value decreases if the new allocation for the ongoing calls in cell k is the same as the previous allocation. The minimization of (1) determines the fittest individual in order to find the optimal channel allocation solution. C. Mutation A mutation rate is selected in order to indicate the probability for a gene in the chromosome to mutate. A low mutation rate is sufficient to prevent any gene in the chromosome to remain fixed to a single value in the population. On the other hand, a high mutation rate will result in random global search for optimal solution. Therefore a moderated value needs to be selected to maintain a balance between such extremes. The parent chromosome is iterated through and randomly determines whether the gene will mutate according to the mutation rate. When the gene which is represented by the channel number is decided to undergo mutation, it will swap the value with the corresponding vector of eligible channels. This process can always produce feasible child chromosome since it does not affect the length of the parent chromosome and does not produce any duplicated channel number. D. Crossover A crossover rate is selected to indicate the probability for parents vectors to crossover to produce a better child chromosome which takes the best characteristics from each of the parents. The proposed crossover strategy is one-point crossover to reduce the computational cost. A single crossover point is selected for both parents vectors. Then the channel numbers which are beyond the crossover point in both vectors are swapped, and results in the child chromosome. V. SIMULATION RESULTS AND DISCUSSIONS In the simulation, the performance of the proposed GA based algorithm for the HCA scheme is evaluated in terms of the blocking probability or dropping probability for the incoming calls. The call blocking probability is calculated by the ratio of the total number of new call blocked and the total number of call arrived in the cellular network system. The performance of the proposed algorithm is compared with HCA scheme which is based on deterministic method, where the channel allocation results are always the same at each simulation without the optimization by GA. The HCA scheme with deterministic method is based on channelordering property, where the first channel in the set of eligible channels is given the highest priority to be assigned to new call request. The representative ratios applied in the algorithm are 21:49 (21 channels in fixed set and 49 channels in the 514 2011 11th International Conference on Hybrid Intelligent Systems (HIS)

dynamic set), 35:35, and 49:21. An example of a valid assignment of channels which fulfills the EMC constraints for the 21:49 representative ratio of the HCA scheme is shown in Fig. 2. This simulation result is optimized by GA and run under nonuniform call traffic distribution using Table II as the initial rate. Fig. 3 shows the call blocking probability result under nonuniform call traffic distribution according to Table II as the initial traffic rates. On the other hand, Fig. 4 shows the call blocking probability performance under uniform traffic distribution with average 15 calls per minute as the initial traffic rate. In the simulation results, the percentage increase of traffic load implies that the traffic rates for each of the cell increased by a percentage with respect to the initial traffic rates. The increase is ranging from a percentage of 0 to 60, with respect to the initial traffic rates. According to Fig. 3, under nonuniform traffic distribution, the lowest call blocking probability is obtained by the 21:49 HCA-GA scheme compared with the other representative ratios as well as the HCA scheme based on deterministic method without GA optimization. The same observation can be obtained from Fig. 4, where the lowest blocking probability is obtained by the 21:49 HCA-GA scheme under uniform traffic distribution. However the 49:21 HCA-GA consumes the shortest running time due to most of its channels are allocated permanently and hence save the time on searching for global minimum on the dynamic set of the channels. Besides that, the characteristics of the GA-based HCA scheme is illustrated in Fig. 5 for the representative ratios of 49:21, 35:35 and 21:49 respectively. According to Fig. 5, the smallest average number of generations required to converge refer to 49:21, which is approximately 5 generations as compared to the 35:35 and 21:49 schemes which require approximately 6 and 7 generations respectively. Therefore the 35:35 HCA-GA scheme can be used to maintain a balance between the extremes of call blocking probability and computational effort. Channel Number 50 45 40 35 30 25 20 15 10 5 Call Blocking Probability 0.9 0.8 0.7 0.6 0.5 0.4 0.3 21:49 HCA-GA 35:35 HCA-GA 49:21 HCA-GA 21:49 HCA 35:35 HCA 49:21 HCA 0 0 5 10 15 20 25 30 35 40 45 50 Cell Number Figure 2. A channel assignment result for the DCA allocation matrix with 21:49 representative ratio under nonuniform traffic distribution at iteration=25. 0.2 0 10 20 30 40 50 60 Percentage Increase of Traffic Load Figure 4. Call blocking performance of HCA-GA for the cellular network with uniform traffic distribution and comparison with the deterministic channel allocation schemes. Call Blocking Probability 0.8 0.7 0.6 0.5 0.4 0.3 21:49 HCA-GA 35:35 HCA-GA 49:21 HCA-GA 21:49 HCA 35:35 HCA 49:21 HCA Fitness function 0-0.2-0.4-0.6-0.8-1 49:21 HCA-GA 35:35 HCA-GA 21:49 HCA-GA 0.2-1.2 0.1 0 10 20 30 40 50 60 Percentage Increase of Traffic Load Figure 3. Call blocking performance of HCA-GA for the cellular network with nonuniform traffic distribution and comparison with the deterministic channel allocation schemes. -1.4 1 2 3 4 5 6 7 8 9 10 Number of generations Figure 5. Characteristics of GA-based HCA scheme for the representative ratios of 49:21, 35:35 and 21:49. 2011 11th International Conference on Hybrid Intelligent Systems (HIS) 515

It can be concluded that when the proposed GA based approach is applied to the HCA scheme, the simulation results in Fig. 3 and Fig. 4 show that for all the representative ratios, GA-based HCA scheme always shows better performance in terms of call blocking probability, under both uniform and nonuniform traffic distribution conditions. Among all the representative ratios, the best performance in terms of call blocking is obtained by the 21:49 HCA-GA scheme. However, in terms of the performance of computation time, 49-21 HCA-GA scheme is faster. VI. CONCLUSION An optimization algorithm based on GA is proposed to solve the NP-completed CAP1 in a cellular mobile network, with better performance in terms of call blocking probability compared to the other HCA scheme based on deterministic method. It can mimic the evolutionary process in nature to optimize the channel assignment problem. GA consists characteristics to evolve through generations and to select the fittest chromosome ensure that it can be self-optimized. The concept of channel reuse scheme avoids the allocation of channels which would cause CCC interference. Hence the computation time to determine this type of interference in the process of channel allocation is reduced. Besides, the combination of the integer genetic representation, the mutation operator and the crossover operator guarantees that the optimum solution can always be found. In addition, the proposed HCA scheme which consists the fixed set of channels is created by equally distributing the channels to all the cells at different reuse scheme. In future, an evolutionary optimization method can be used to assign the channels from the fixed set based on the call traffic load of the cell. REFERENCES [1] K.A. Smith, and M. Palaniswami, Static and dynamic channel assignment using neural networks, IEEE J. Select. Areas Commun, vol. 15, pp. 238-249, 1997. [2] G. Vidyarthi, A. Ngom, and I. Stojmenovic, A hybrid channel assignment approach using an efficient evolutionary strategy in wireless mobile network, IEEE Trans. Veh. Techno, vol. 54, pp. 1887 1895, 2005. [3] H.G. Sansalidis, P.P. Stavroulakis, and J. Rodriguez-Tellez, An efficient evolutionary algorithm for channel resource management in cellular mobile systems, IEEE Trans. Evol. Comput, vol. 2, pp. 125-137, 1998. [4] J. Chen, S. Olafsson, and X. Gu, Observation on using simulated annealing for dynamic channel allocation in 802.11 WLANs, Proc. VTC Spring, pp. 1801-1805, 2008. [5] C.Y. Lee, and H.G. Kang, Cell planning with capacity extension in mobile communications: a tabu search approach, IEEE Trans. Veh. Technol., vol. 49, pp. 1678-1691, 2000. [6] M.A. C. Lima, A.F.R. Araujo, and A.C. Cesar, Dynamic channel assignment in mobile communications based on genetic algorithms, 13th IEEE International Symposium on Personal, Indoor and Mobile Radio Communication, vol. 5, pp. 2204-2208, 2002. [7] J. Yoshino, and I. Othomo, Study on efficient channel assignment method using the genetic algorithm for mobile communication systems, Journal Soft Computing A Fusion of Foundation, Methodologies and Applications, vol. 9, pp. 143-148, 2005. [8] L.M. San Jose-Reveuelta, A new adaptive genetic algorithm for fixed channel assignment, Journal Information Sciences, vol. 177, pp. 2655-2678, 2007. [9] S.C. Ghosh, B.P. Sinha, and N. Das, Channel assignment using genetic algorithm based on geometric symmetry, IEEE Trans. Veh. Techno., vol. 52, pp. 860-875, 2003. [10] M. Cuppini, A genetic algorithm for channel assignment problems, European Trans. Telecom. Rel. Techn., vol. 5, pp. 285-294, 1994. [11] W.K. Lai, and G.C. Coghill, Channel assignment through evolutionary optimization, European Trans. Telecom. Rel. Techn., vol. 45, pp. 91-96, 1996. ACKNOWLEDGMENT The authors would like to acknowledge the financial assistance of the Ministry of Higher Education of Malaysia (MoHE) under Fundamental Research Grant Schemes (FRGS) no. FRG0104-TK-1/2007. 516 2011 11th International Conference on Hybrid Intelligent Systems (HIS)