I In recent years, regarding to basic progresses in design of - - PDF document

• Vol. 1(3), Oct. 2015, pp. 227-233

227 Article History: JKBEI DOI: 649123/11033 Received Date: 14 Jun. 2015 Accepted Date: 22 Sep. 2015 Available Online: 29 Oct. 2015

Presentation of an Optimum Routing Template in Wireless Sensor Networks (WSNs) Based on Imperialist Competitive Algorithm

Mohammad Gharary, Javad Vahidi Computer MSc student, PHD analysis of algorithms Phone Number: +98-915-6029244

*Corresponding Author's E-mail: Gharary@me.com

Abstract

In recent years, regarding to basic progresses in design of integrated orbits, wireless telecommunication and designing sensor a special type of wireless network named wireless sensor networks (WSNs) has been considered by researchers and different industries. One of the major challenges in wireless sensor networks is routing to send data on the network. In this paper was presented the method to find optimized routes by using imperialist competitive algorithm. In the method, initially in routing phase, we model the state of the network based on a graph, then according to the possible routes we routing. The target functions in the algorithm are to find the lowest available route between sender and recipient data. Two important operators in the proposed method are assimilation and

• revolution. In the paper, assimilation rate has been considered 90% and revolution rate is 50% and the

maximum iteration in algorithm is considered 1000 in the base mode. According to the comparisons were done, the proposed method does the operation routing more efficient compared to methods based

• n genetic algorithm (GA).

Keywords: Routing, Wireless Sensor, Network, Imperialist Competition.

• 1. Introduction

Many usages of wireless sensor networks need to send data sure and to reduce of nodes energy

• consumption. On the other hand, fundamental challenge in the networks is presenting an appropriate

routing protocol, due to natural limitations of wireless sensor networks. In many usages and protocols

• f routing, to reduce energy consumption and following prolongation of networks lifetime are

important, due to sensor nodes energy limitations. In these networks, energy consumption consists of turning on Radio communications, energy necessary to send and receive of date packets types and energy necessary to turning on sensors. Most routing algorithms model problem state space in form

• f the graph. Then they are looking for the optimum route among the solutions available in the graph.

In this paper, in the routing phase in the network there are different routes to send data from sender to recipient. Whatever we can find a route at minimum distance between sender and recipient, we have been done routing better. Since the distance between the nodes in the graph of network makes up the different route; one way is that first calculate cost of all routs then decide to choose optimum

• route. But this need high time due to number of network present nodes and we can say the algorithm

is a factorial stage. Another way is Production of some routes random and performance of an

• ptimization operation by optimum algorithms. So, optimum criterion is here the distance between

present nodes and routes. Then in the paper, using of imperialist competitive algorithm, we are looking for a method to find minimum possible distance between node of sender and recipient data.

I

Mohammad Gharary et al. / Vol. 1(3) Oct. 2015, pp. 229-235 JKBEI DOI: 649123/11033 228

Journal of Knowledge-Based Engineering and Innovation (JKBEI) Universal Scientific Organization, http://www.aeuso.org/jkbei ISSN: 2413-6794 (Online)

• 2. The Work Done

Rein Form protocol [7] has been presented to form beginning to end, in order to guarantee packet transmission with reliability requires. The protocol is considering the importance and reliability to its packet transmission, it sends multiple copies of the transmitted packet to some neighbor nodes. Number of necessary routes to reliability ensures associated transmission packet is calculated based

• n transmission channel error and network topologic information (number of step from source to

destination). The protocol isn’t considered energy parameter. EQSR protocol [13], energy parameters of the sensor nodes, the ratio of signal power to the noise power of the communication link and empty buffer of next node, in order to appropriate route choose and send data to sink. It uses multi-path routing transmission to increase delivery packets to the

• destination. To determine multiple routes between source and destination, sink determines the

required routes by sending route request packet (RR) to source node. EARQ protocol [13] immediately sends information packets to sink with high reliability. The protocol to appropriate route choose, considers neighbor nodes energy cost. To send immediate packets choose node that have less latency. The protocol not discriminates to send packets in reliability and

• latency. And no priority is not considering in send different packets.

PAP protocol [18] uses geographic send method greedy. The protocol considers the packet required speed in send to immediate packets; i.e. it choose nodes as next node which can supplies acquired speed of packet. Notable limitation of RAP protocol is that sends to geographic greedy is not considers network local conditions such as load balancing, level of congestion and communication channel

• quality. Therefore, RAP routing-path protocol has been unpredictable delay to every stage in dynamic

environment of wireless sensor network that influence the network effectiveness and timely and successful send ensure of packet. Also other weakness of RAP algorithm is that is not considers the nodes present energy in route. MCMP protocol [20] considers reliability and the communication link delay as routing-path parameters to send packets. And it sends data to increase reliability for multi-path routing and by solving linear programming problem. The protocol is not considering neighbor nodes parameters and send spread speed. Packets are sent toward the congestion nodes in this protocol. For this reason, number of sensitive packets reduces to delivered time in destination due to overcrowding of nodes. MMSpeed protocol [22] was completed by speed protocol, and it is a multi-path routing protocol and multi-track. The protocol is considering the reliability parameters and time delay to send packet. A weakness of MMspeed protocol is that has not been considered energy important parameter to

• ptimum route chooses.
• 3. Proposed Method

In the part, we will review and introduce proposed algorithm to wireless sensor networks routing. We can model the routing problem in the form of a graph. In this case, any of nodes will convert graph

• tops. And any of the manes convert to present routes in the graph. Used methods in this work are

based on imperialist competitive algorithm. In the algorithm concepts of country, imperialistic, colony, revolution and etc. are fundamental concepts. 3.1 Problem Model We can model a routing problem or a space-based sensor network by a graph. The aim is to find the shortest time possible to send data over the network from source to destination. In the graph above assuming that the node source is N1 and node data receive destination is N11; can be considered data transport cost of graph nodes by matrix to below figure. Cost= [100 101 102 98 97 104 96 101 100 103 98 97 99 101 100 103]

Mohammad Gharary et al. / Vol. 1(3) Oct. 2015, pp. 229-235 JKBEI DOI: 649123/11033 229

Journal of Knowledge-Based Engineering and Innovation (JKBEI) Universal Scientific Organization, http://www.aeuso.org/jkbei ISSN: 2413-6794 (Online)

Present manes between graph nodes from N1 to N11 enter in the matrix. And we cost the value of each mane due to data transmission time between nodes by these manes. Under consideration above graph can consider three classes between graph nodes from N1 to N11. And consider present and usable transmission routes in each class. For example, no data is unlikely to node N9 unless passed the N5 or N6. Actually, the existence of class ensures a correct transmission from possible routes between

• nodes. After the stage, we will draw the routes to the graphic form. Cost total function is equal costs
• f each route (transmission time between nodes) from N1 to N11 so that pass all class and regard

correct transmission. In the next step of the proposed method, colony country has moved to size unit x in direction of line of colony to imperialist and trend new position. X is a random number with same distribution (or each other appropriate distribution). Since the algorithm might get in local optimums therefore need revolution procedure. In imperialist competitive algorithm, revolution is modeled by random movement of a colony country into a new random position. Revolution in algorithm approach makes totality of evolutionary motion is saved the getting in optimum local valleys so in some instances makes to improve a country position and take it a more optimum limited area [6]. These stages continue until is zero the convergence error or number of algorithm iteration to reach the specified number.

• 4. Data Analysis

In this part we describe performance of the proposed method to wireless sensor network routing and will review research findings and experience results. Function off proposed cost is calculated below figure. Firstly, You conser arrays of start, finish and cost:

Start [1 1 1 2 2 3 3 4 4] Finish [2 3 4 5 6 6 7 7 8] Cost [150 100 120 140 100 125 150 200 230]

In the above arrays, for example, we need to send data from node 1 to node 3, the cost equivalent 100 and need to send from node 3 to node 6 the cost equivalent 125 and so on need to continue to

• ther costs base on cost matrix and start and finish. Actually, each existing route is between start node

and finish node of a country. The aim is to find a country that is consists of route start to finish and have the lowest cost and according to expressed concepts introduce some countries that every country represents a route and a cost function. To performance of method firstly introduce some country. These numbers are read by user from the input. Whatever number of countries is high; time to reach convergent is high and answer is more

• accurate. It should be noted that the number of countries must is less than or equal to the total number
• f graphs routes. The aim is to reach to a collection of country named imperialist. For example, number
• f 40 countries is proposed. Our proposed system chooses 40 routes the form of random then

calculates the cost function for each route. Therefore, cost function matrix that is matrix 40*1 sort

• descending. And the last element is considered as the imperialist and the rest of the 39 element as
• colonial. The present should is performed tow important action of imperialist competitive algorithm;

first action, is colonial nearing to imperialist. The action takes place by numeral average catch. Numeral average can obtain by below relevance: By performance of the action countries precede toward the best cost function and the action perform so far that we reach one of the problem end condition. May be get in some performances on the moving colonies toward countries imperialist in the best local cost function and algorithm is

• stopped. To get out of this jam must take place second action of imperialist competitive algorithm

Mohammad Gharary et al. / Vol. 1(3) Oct. 2015, pp. 229-235 JKBEI DOI: 649123/11033 230

Journal of Knowledge-Based Engineering and Innovation (JKBEI) Universal Scientific Organization, http://www.aeuso.org/jkbei ISSN: 2413-6794 (Online)

named revolution. To perform of revolution operation, we define a random number for each of the

• classes. The meaning of the classes is nodes between source node and destination node. However,

node that is same level in problem graph, it was put in a class. We know each class consist of node with number of level. This number to first class is 1 or 2; to second class should be 1, 2 or 3 and to third class are 1 or 2 values that is introduced the form of random. Then introduce the other random number until it choose the class between 1 to 3 and take place revolution in the existing nodes of the class. In a class that is a candidate for revolution one of the produced random nodes in before stage will be replaced to another random node in the class. This occurs for half of countries (routes) i.e. occur for 20 countries. So revolution percentage is 50%. If the percentage is higher, may be to reach an answer is longer caused lack faster convergence and if is low, the algorithm stick. After performance of revolution in all countries, all countries updated and again go back to the beginning of the algorithm and is calculated their cost function and again the cost function matrix is re-arranged and its fortieth element that actually has the lowest cost as imperialist will be selected and the rest of countries (routs) as colonial are considered and the trend continues. Notable note is that there isn’t any reason a country or route which is selected as imperialist in a repeat, is selected as imperialist in the next iterations. Proposed algorithm firstly performed in the 20 countries (routes). Cost function and route figure is visible in the last repletion in figure below:

Figure 1: Cost function for 20 countries

In figure the algorithm reached to end of his work for 20 countries and it has been found appropriate route to send data from source node to destination node. In table of 4-1 the perfect information is shown for all the passes. As specified in the table, after to passing 16, amount of cost has been fixed and never changes. Actually, algorithm reached to exact answer in this passing.

Table 1: The complete information for all passes for 20 countries

Number of Passing Cost Number of Passing Cost 1 2 3 4 5 6 7 8 9 10 1307 1283 1157 1157 1051 1051 1026 1024 1024 924 11 12 13 14 15 16 17 18 19 20 924 919 919 919 900 900 769 769 769 769

Mohammad Gharary et al. / Vol. 1(3) Oct. 2015, pp. 229-235 JKBEI DOI: 649123/11033 231

Journal of Knowledge-Based Engineering and Innovation (JKBEI) Universal Scientific Organization, http://www.aeuso.org/jkbei ISSN: 2413-6794 (Online)

Proposed algorithm will be stopped in the below two statuses A) If the number of iterations reaches to amount of maximum iteration, this will vary and is changeable or adjustable by user. B) The convergence error amount is equal to number that is shows the amount of distance of total countries and it is determining factor to observation of the subject that whether converted countries to imperator? 4.1 To compare proposed method to Genetic Algorithm In the table 4-2, proposed method information is presented to 20 countries. The results with its amount are comparable in GA.

Table 4-2: the complete information for all passes for 20 countries in ICA & GA

Number

• f Passing

Cost of Proposed method Cost of Genetic Number of Passing Cost of proposed method Cost of Genetic 1 2 3 4 5 6 7 8 9 10 1307 1283 1157 1157 1051 1051 1026 1024 1024 924 1328 1296 1169 1176 1068 1069 1045 1046 1045 952 11 12 13 14 15 16 17 18 19 20 924 919 919 919 900 900 769 769 769 769 939 930 931 930 925 915 783 789 792 790 In graph 4-1, to compare of proposed method for the 20 countries is visible in the proposed method by GA method. As you can see in the graph, the ICA method is more optimal than the GA.

Figure 2: Comparison Proposed method with GA for 20 countries

Also in the table 4-3, proposed method information has been presented for 100 countries. The results with its amount are comparable in GA.

Mohammad Gharary et al. / Vol. 1(3) Oct. 2015, pp. 229-235 JKBEI DOI: 649123/11033 232

Journal of Knowledge-Based Engineering and Innovation (JKBEI) Universal Scientific Organization, http://www.aeuso.org/jkbei ISSN: 2413-6794 (Online)

Table 3: the complete information for all passes for 100 countries

Passing ICA GA Passing ICA GA Passing of 1 to 9 1307 1359 Passing of 43 to 48 1026 1055 Passing of 10 to 13 1283 1325 Passing of 49 to 57 1024 1053 Passing of 14 to 18 1206 1239 Passing of 58 to 61 924 978 Passing of 19 to 23 1166 1200 Passing of 62 to 72 919 993 Passing of 24 to 34 1157 1189 Passing of 73 to 81 900 976 Passing of 35 to 42 1051 1089 Passing of 82 to end 769 843 Also in figure below, there is the graph of comparison proposed method with GA for 100 countries:

Figure 3: Comparison proposed method with GA for 100 countries

Conclusion

In this paper, an efficient method and accurate has been presented in order to routing data in wireless sensor network. Procedure of proposed algorithm is that firstly some route from source to destination node named country the entrance received and calculated the cost function of each route and created the cost matrix of the number country then sorted the matrix descending and the last element of the matrix that had the lowest cost among the countries; consider as imperialist and the rest of countries named colonial. Then the colonials in each iteration move toward imperialist and function values of each country and also the updated cost matrix and new imperialist choose. And the

• peration will be continuing until reach the one of the algorithm end condition. To escape the local
• ptimum positions that in the algorithm are called being stuck; is used revolution procedure. The

revolution rate is 50% and the algorithm iteration maximum is 1000 in base. The comparisons were made; proposed method compared to the GA method does more optimum the routing operation.

200 400 600 800 1000 1200 1400 1600 ICA GA

Mohammad Gharary et al. / Vol. 1(3) Oct. 2015, pp. 229-235 JKBEI DOI: 649123/11033 233

Journal of Knowledge-Based Engineering and Innovation (JKBEI) Universal Scientific Organization, http://www.aeuso.org/jkbei ISSN: 2413-6794 (Online)

References

[1] A. Razzaque and M. M. Alam, “Multi-Constrained QoS Geographic Routing for Heterogeneous Traffic in wireless Sensor Networks,” IEICE TRANS. COMMUN., vol. E 91-B, no. 8, pp. 2589-2601, 2008 [2] B Deb, S. Bhatnagar, and B. Nath, “ReInForM: Reliable information forwarding using multiple paths in sensor networks,”in 28th Annual IEEE International Conference on Local Computer Networks, 2003. LCN’03. Proceedings, 2003,

• pp. 406-415

[3] H. Zhou, Y. Wu, Y. Hu, and G. Xie, “A novel stable selection and reliable transmission protocol for clustered heterogeneous wireless sensor networks,” Computer Communications, vol. 33, no. 15, pp. 1843-1849, 2010. [4] J. Kulik and W. Heinzelman, “Negotiation-Based Protocols for Disseminating Information in,” Wireless Networks, pp.

169-185, 2002.

[5] Liang, Junbin, Ming Liu, and Xiaoyan Kui. "A survey of coverage problems in wireless sensor networks." Sensors & Transducers (1726-5479) 163.1 (2014). [6] P. Kok, K. Loh, H. W. Jing, and Y. Pan, “Performance Evaluation of Efficient and Reliable Routing Protocols for Fixed- Power Sensor Networks,” IEEE Trans. on wireless communication, vol. 8, no. 5, pp. 2328-2335, 2009. [7] Rawat, Priyanka, et al. "Wireless sensor networks: a survey on recent developments and potential synergies." The Journal of supercomputing 68.1 (2014): 1-48. [8] S. Lee, B. Bhattacharjee, S. Banerjee, and B. Han, “A general framework for efficient geographic routing in wireless networks,” Computer Networks, vol. 54, no. 5, pp. 844-861, 2010. [9] S. Subramanian and S. Shakkottai, “Geographic Routing with Limited Information in Sensor Networks,” IEEE Trans, on Information theory, vol. 56, no. 9, pp. 4506-4519, 2010. [10] Y.-chung Fan, A. L. P. Chen, S. Member, and I. C. Society, “Efficient and Robust Schemes for Sensor Data Aggregation Based on Linear Counting,” IEEE Trans, on parallel and distributed systems, vol. 21, no. 11, pp. 1675-1691, 2010. [11] Yang, Kun. Wireless Sensor Networks. Springer, 2014.