Dynamic Energy Aware Gur Game based Algorithms for Self Optimizing Wireless Sensor Networks

This paper presents, Application of Gur Game Based Algorithm on Wireless Sensor Networks (WSNs) deployed to monitor Homogenous and Heterogeneous Grid in order to achieve Quality of Service (QoS) = 0.40 and 0.50. Further, the objectives of all these algorithms are to maximize the coverage of the sensor area while conserving energy consumed by sensor nodes. This is achieved via carefully activating/deactivating the sensors while maximizing the coverage area


INTRODUCTION AND MOTIVATION
A WSN consists of spatially distributed autonomous sensors to monitor physical or environmental conditions, such as temperature, sound, vibration, pressure, motion or pollutants [1][2][3] and to cooperatively pass their data through the network to a main location. The development of WSNs was motivated by military applications such as Battlefield Surveillance; today such networks are used in many industrial and consumer application, such as Environmental Monitoring, Habitat Monitoring, Acoustic Detection, Seismic Detection, Military Surveillance, Inventory Tracking, Medical Monitoring, Smart Spaces and Process Monitoring [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16]. Sensors are low-powered devices with very little computational capability. WSNs allow sensors to work together in order to cover wide areas comprehensively. In a typical application, a WSN is scattered in a region where it is meant to collect data through its sensor nodes. The WSNs have many limitations such as Hostile Environment, Random topology, Limited Resources, Design Challenges. In this paper, we are specifically talking about the Random Distribution of the sensors in the Homogeneous and Heterogeneous Grids and Power Restrictions (we want to minimize the energy dissipation as much as possible). In order to handle these two open issues we have taken the help of mathematical Gur Game Based Algorithms and its application on WSNs [15,16].

PROBLEM DESCRIPTION AND CONTRIBUTIONS
In this paper, we have shown the Application of Gur Game on WSNs. More specifically we have shown the (1) Application of Generalized Gur Game based Algorithm on WSNs monitoring 4 x 4 Homogenous Grid to achieve QoS = 0.40 and 0.50. This approach is Application-Specific. (2) Application of Gur Game based Algorithm on WSNs monitoring 4 x 4 Heterogeneous Grid to achieve QoS = 0.40 and 0.50. This approach is Application-Specific and cannot be generalized because of the degree of Heterogeneity, is extremely high.
For our simulation, purpose we have used Java technology moreover, Gur Game is a random approach and can be efficiently simulated with Java Technology. All simulation results are validated through MATLAB.
The rest of the paper is organized as follows. Section 2 describes some of the previous work already done in this area in terms of energy aware algorithms for WSNs. Section 3 describes the Application of Gur Game Based Algorithms on WSNs (Generalized, monitoring Homogenous and Heterogeneous Grid) is supported by the Simulations results from MATLAB and Java Technology and Limitations for Homogenous and Heterogeneous WSNs Design, followed by Section 4 concludes the paper followed by the references.

APPLICATION OF GUR GAME BASED ALGORITHMS ON WIRELESS SENSOR NETWORKS
A random and greedy mathematical paradigm of the Gur Game has been proposed in [6,7]. This technique allow the nodes to communicate with the base station at a regular interval of time, and base station send feedback in order to manage nodes on basis of packets received. The result is a robust sensor network that allows the base station to dynamically adjust the resolution of a network based on feedback it receives from the sensors in the network. Recently the authors of [15,16] introduced self optimizing algorithms to regulate the process of activating sensors while maximizing the number regions covered by sensor nodes. They had proposed a dynamic clustering algorithm that employs the concept of connected dominating sets. They also improved the earlier Ants Algorithm and Genetic Algorithm to take into consideration the dynamic nature of WSNs.
In this paper, we report the results obtained by implementing Gur Game Based Algorithms. Gur Game is a mathematical algorithm, was originally introduced by Brian Tung and L. Kleinrock in [6,7]. Instead of using number of active nodes, we use ratio between coverage area and number of active nodes as QoS parameter. We have learnt the application of Gur Game based algorithm on WSNs and the reward function for simulation purposes from the authors of the paper [15,16] however; we have taken a different approach to design the algorithm and its application on the WSNs.

Simulation Scenario 1
In order to achieve the QoS = 0.40 (from a Homogenous 4 x 4 Grid having 16 regions) we have should have 16 regions and these 16 regions must be monitored by the 40 sensors as we know that 16/40 = 0.40 and if we take this as a vice versa we will have 40/16 = 2.5 i.e. each region is monitored by the 2.5 sensors. Further, in order to achieve the QoS = 0.40 (from a Heterogeneous 4 x 4 Grid having 15 regions) we have should have 15 regions and these 15 regions must be monitored by the 37.5 sensors and It is not possible physically and technically. For getting X = 0.40 we have X = 37.5/15 = 2.5 or X = 15/37.5 = 0.40, where 37.5 is the active number of sensors in the 15 regions. Briefly, we cannot achieve 37.5 active sensors for the design we have chosen for our simulations. Therefore, to tackle this problem we have used the round off code (explained in Section-3.6.2) to round off the 37.5 as 37 and just saving nearly 1 more sensor to get active.

Simulation Scenario 2
In order to achieve the QoS = 0.50 (from a Grid having 16 regions) we have should have 16 regions and these 16 regions must be monitored by the 32 sensors as we know that 16/32 = 0.50 and if we take this as a vice versa we will have 32/16 = 2 i.e. each region is monitored by the 2 sensors. Further, in order to achieve the QoS = 0.50 (from a Grid having 15 regions) we have should have 15 regions and these 15 regions must be monitored by the 30 sensors as we know that 15/30 = 0.50 and if we take this as a vice versa we will have 30/15 = 2 i.e. each region is monitored by the 2 sensors. 778 | P a g e w w w . i j c t o n l i n e . c o m

Testbed, Experimental Setup and Simulation Outputs
We have used Java Technology (i.e. JDK 1.6) for simulations and MATLAB for drawing the Graphs and these two software are running on top of the IBM System x, running with Novell's SUSE Linux Enterprise Server 11. We have used the java.util.Random class allows you to create objects that produce pseudo-random numbers with uniform or gaussian distributions according to a linear congruential formula with a 48-bit seed. To handle the Redundancy we have we have used HashSet method to create a HashSet, which allows no duplicates but in order to ensure correct order, since HashSet does not we have to use Collections method. Step3: If Sensor no. < total Sensors goto Step4 Else goto Step10.

Application of Gur
Step4: Enter If Sensor is active or not.
Step10: Count the number of active Sensors.
Step11: Calculate total Regions covered by active Sensors (with repetition).
Step13: Calculate the value of X which is the ratio between Regions covered (result of Step12) and active Sensor nodes.
Step16: If Sensor was active, assign the value of Reward r Else Penalized with Probability 1-r. step17: Compare X with 0.40, If X=0.40, QoS is achieved, which implies that each Region is covered by more than 2.5 active Sensor nodes on an average Else QoS is not achieved.

Or
Compare X with 0.50, If X=0.50, QoS is achieved, which implies that each Region is covered by more than 2.0 active Sensor nodes on an average Else QoS is not achieved.    Step2: Enter the number of Regions[0...14] the field is divided into.
Step4: Enter If Sensor is active or not.
Step5: If Sensor is active goto Step6 Else goto Step11.
Step9: If Sensor index is {5} then total Regions covered is 7 Else if Sensor index is {6,9,10} total Regions covered is 8.
Step10: Regions covered by the Sensor are designated to the active Sensor according to the adjacent Region for every node.
Step12: Count the number of active Sensors.
Step13: Calculate total Regions covered by active Sensors (with repetition). w w w . i j c t o n l i n e . c o m Step14: Remove duplicates from Step13.
Step15: Calculate the value of X which is the ratio between Regions covered (result of Step14) and active Sensor nodes.
Step18: If Sensor was active, assign the value of Reward r Else Penalized with Probability 1-r.
Step19: Compare X with 0.40, If X=0.40, QoS is achieved, which implies that each Region is covered by more than 2.5 active Sensor nodes on an average Else QoS is not achieved.

Or
Compare X with 0.50, If X=0.50, QoS is achieved, which implies that each Region is covered by more than 2.0 active Sensor nodes on an average Else QoS is not achieved.

Homogeneous Grid
The simulations results achieved by the application of Gur Game based Algorithms on 4 x 4 Homogenous Grid presented in figure (3)(4)(5)(6)(7)(8)(9) shows that the number of sensors to be activated in order to achieve the QoS = 0.40 is same as the number of sensors needed to activate and to achieve the QoS of 0.50. Therefore, we are achieving high QoS = 0.50 in comparison to QoS = 0.40 at least expense of energy dissipation and number of sensors.

Heterogeneous Grid and its Limitations
In order to achieve the QoS of 0.40 we have to activate the sensors and control will be in the user's hands as it is Application-Specific. The number of sensors to be activated to monitor the Homogenous Grid is 9 and the number of sensors to be activated to monitor the Heterogeneous Grid is 5 only so we are saving lots of sensors and achieving high QoS. For getting X = 0.40 the story is difficult as X = 37.5/15 = 2.5 or X = 15/37.5 = 0.40, where 37.5 is the active number of sensors in the 15 regions and this is not possible during the coding and for the design we have chosen. Briefly, we cannot achieve 37.5 active sensors. Therefore, to tackle this problem we have used the following code to roundoff the 37.5 as 37 and just saving nearly one more sensor to get active. System.out.println("x="+X); _____________________________________________________________________________ Therefore, after this code appears we have 37.5/15 ~ 37/15 = 0.40 and hence achieving the QoS. We have to use code to make our assumptions to run otherwise we cannot do anything for achieving the X = 0.40.
In this paper, we introduce self optimizing algorithms to regulate the process of activating sensors while maximizing the number regions covered by sensor nodes. Specifically, we have used Gur Game based Algorithms and its Application on WSNs for 4 x 4 Homogenous and Heterogeneous Grid to achieve the QoS = 0.40 and 0.50. d.
The simulations results achieved by the application of Gur Game based Algorithms on 4 x 4 Heterogeneous Grid presented in figure (10)(11)(12)(13)(14)(15)(16)(17) shows that the number of sensors to be activated in order to achieve the QoS = 0.40 which is very high in comparison to the number of sensors needed to activate and to achieve the QoS = 0.50. Therefore, we are achieving high QoS = 0.50 in comparison to QoS = 0.40 at very least expense of energy dissipation and number of sensors. The design presented is here is only for 4 x 4 Heterogeneous architecture i.e. Application-Specific or in other words we have only explored the Application of Gur Game based Algorithm on WSNs monitoring 4 x 4 Heterogeneous Grid to achieve QoS = 0.40 and 0.50. This approach is Application-Specific and cannot be generalized, as the degree of Heterogeneity, is extremely high.

CONCLUSION
In this paper, we introduce self optimizing algorithms to regulate the process of activating sensors while maximizing the number regions covered by sensor nodes. Specifically, we have used Gur Game based Algorithms and its Application on WSNs for 4 x 4 Homogenous and Heterogeneous Grid to achieve the QoS = 0.40 and 0.50.
The simulations results achieved by the application of Gur Game based Algorithms on 4 x 4 Homogenous Grid presented in figure (3)(4)(5)(6)(7)(8)(9) shows that the number of sensors to be activated in order to achieve the QoS = 0.40 is same as the number of sensors needed to activate and to achieve the QoS of 0.50. Therefore, we are achieving high QoS = 0.50 in comparison to QoS = 0.40 at least expense of energy dissipation and number of sensors.
The simulations results achieved by the application of Gur Game based Algorithms on 4 x 4 Heterogeneous Grid presented in figure (10)(11)(12)(13)(14)(15)(16)(17) shows that the number of sensors to be activated in order to achieve the QoS = 0.40 is very high in comparison to the number of sensors needed to activate and to achieve the QoS of 0.50. Therefore, we are