Modeling analysis and simulation of scalability in wireless sensor networks with capacity criterion

Master's Thesis in Telecommunication-Systems

Abstract

This research is about the issue of scalability in wireless sensor networks with imaging capability, which by considering a relatively practical scenario of the sensor network, and based on performance criteria of outage capacity and ergodic capacity of the network, the scalability has been analyzed, mathematically modeled and simulated.

Scalability is basically to determine The effects of increasing or decreasing the number of sensors in the desired parameter, when the goal is to prevent high costs of network construction due to unnecessary increase in the number of sensors or excessive reduction of network efficiency due to the lack of density of the number of sensors. To achieve this goal, we first tried to use a scenario that uses network clustering. First, by obtaining the distance distributions for the sensors with each other and with the cluster head (header) and also the distribution of the distance of the cluster heads to the main station, through simulation, we were able to obtain the interference distribution and, as a result, the distribution of the signal-to-noise ratio and interference in this network, for both intra-cluster and extra-cluster layers. network, we have obtained to estimate the capacity distribution function. By comparing these analytical results with the results of the simulations, we confirmed their correctness.

To examine the issue of scalability with the help of simulations for different network parameters, such as changing the density of sensors and cluster heads at the network level, changing the values ??of the channel attenuation profile, changing the threshold for disconnection capacity, adding feeding (with Rayleigh distribution) and so on. We have obtained and compared the average and total values ??of ergadic capacity and cutoff in each case. Starting from the sparse (thin) scenario and moving towards the dense (dense) scenario, first the total capacity values ??will increase with the addition of the number of sensors and the increase of the received power in the cluster heads, but with the continuation of this process, the interference effects will prevail over these factors and the amount of capacities will start to decrease after reaching the maximum points, which is considered as the optimal limit of densities. Also, for probability distributions of capacity and ratio of signal to noise and interference, in both layers of the network, analytical formulas or their approximations have been presented, and based on this, an attempt has been made to obtain optimal points for the density of sensors and cluster heads.

Key words: wireless sensor network, ergadic capacity and disconnection, clustering, scalability.

Introduction

Networks Sensory is one of the most widely used topics in collecting information from the environment around human life. Not only checking the surrounding environment but also controlling it is one of the applications of sensor networks. In the case that the sensors are wireless, in addition to increasing the complexity, many applications and benefits are also obtained.

This network consists of hundreds or thousands of sensors and each sensor is called a node. In a wired system, installing many sensors is usually not possible due to the increased cost of wiring. Places that were previously inaccessible, such as the inside of factory engines, as well as dangerous and moving places, can be accessed by WSNs [14]. Structurally, wireless networks can be divided into two general categories. They are a group with a cellular structure [1] where all the nodes are directly connected with the central station [2] and sending and receiving information from one node to another node is done and controlled through this central station. The second category is networks with ad hoc structure (case, ad hoc). In these networks, all nodes are the same in terms of their abilities and tasks. In this way, ad hoc wireless networks consist of a number of nodes that are able to automatically form a network without the need for a central station or any pre-built infrastructure equipment [6]. Due to this feature, ad hoc wireless networks have the ability to be formed and set up quickly using existing nodes..

Wireless sensor networks, which are a subset of contingency networks, are slightly different from it; It is necessary to finally transfer all the information collected from the environment to a central station, similar to the methods of routing and relaying information in contingency networks. However, unlike cellular networks, this central station does not play a role in how the network is produced and controls the routing and communication between nodes or sensors [26]. In a wireless sensor network, in addition to the task of digitizing the received information, each sensor has the task of performing some light processing on the collected data and also has the ability to communicate with other nodes; in this way, each node acts as a source and as a data transmitter. And if necessary, it sends the data sent from other nodes to the destination. This method is different from networks with a cellular structure, where all nodes are connected only to the central station, and the communication of each node with another node is done by it. It is not necessary that the position of these sensors is determined in advance, but the location of these sensors can be chosen randomly[11].

Nodes in WSN are distributed like an Ad-Hoc network and are inactive for most of their time, but they are suddenly activated when they detect an event. When the sensors observe the desired phenomenon, the desired event is reported to a central station.

Various applications and the random state of wireless sensor networks have created many challenges for research in this field, for example: media access control [4], routing protocols, power control and scalability (the ability to change the size of the network and the density of sensors) [11].

Uncertain topology and channel changes over time, many problems has created in the field of research, design and implementation of these networks. Such changes are routing algorithms [5], network efficiency estimation [6], network coverage [7], expandability and scalability [8] and . has faced certain complications that are the subject of today's common research [10, 24, 25, 26].  The need to reduce the complexity of wireless sensor networks has prompted researchers to remove limiting resources, such as the limitation of the communication channel in the domains of time, frequency and code, energy and life span limitations, and so on. , use it in the most appropriate way and move towards the optimal sharing of resources.

Different applications for all types of sensors play a decisive role in determining the priorities of sharing resources; For example, when only temperature or humidity control sensors are used, the amount of information will be small, and therefore increasing the lifetime of the network is the first priority and reducing the amount of interfering information is the second priority. In other applications, such as military applications, for comprehensive control of the environment, there is a need to establish much more cooperation between sensors, and as a result, parameters such as information relay and routing will have more priority. In medical applications, the accuracy of information collection needs to be high, and on the other hand, in an application such as magnetic resonance imaging[9], achieving this high accuracy requires collecting a large amount of information, and the issues of compression and aggregation of information will be discussed; And finally, for applications such as imaging, where the amount of information is high, but compared to medical applications, the size of the area covered by the network is much larger, in addition to high accuracy, it is necessary to pay attention to positioning, controlling the way information is collected, how to cooperate between sensors, both for collecting and sending information; Especially, the required number of sensors per unit area or the density of sensors plays a decisive role in this. If the number of sensors is too high, although the accuracy increases, but in addition to increasing network costs, factors such as interference and how to route reduce its efficiency. On the other hand, too little number of sensors will cause a crisis in the possibility of monitoring the network and collecting complete information.

In order to increase the accuracy and get closer to the reality, the amount of data should be increased. For this purpose, the number of network nodes should be increased. By increasing the number of nodes of a sensor network per unit area, in addition to increasing the ratio of signal to noise and interference or reducing the bit error rate [10] (BER) due to the increase in power, the power of interference signals reaching each node also increases.  Reaching the optimal point where the ratio of signal to noise and interference or BER is minimized is one of our goals in this project, which is proposed along with the capacity criterion.