Noisy distance measurements are a pervasive problem in localization in wireless sensor networks. Neural networks are not commonly used in localization, however, our experiments in this paper indicate neural networks are a viable option for solving localization problems. In this paper we qualitatively compare the performance of three different families of neural networks: Multi-Layer Perceptron (MLP), Radial Basis Function (RBF), and Recurrent Neural Networks (RNN). The performance of these networks will also be compared with two variants of the Kalman Filter which are traditionally used for localization. The resource requirements in term of computational and memory resources will also be compared. In this paper, we show that the RBF neural network has the best accuracy in localizing, however it also has the worst computational and memory resource requirements. The MLP neural network, on the other hand, has the best computational and memory resource requirements.

Abstract-Energy

Localization arises repeatedly in many location-aware applications such as navigation, autonomous robotic movement, and asset tracking. Analytical localization methods include triangulation and trilateration. Triangulation uses angles, distances, and trigonometric relationships to locate an object. Trilateration, on the other hand, uses only distance measurements to identify the position of the target. Figure 1A, describes a simple example of trilateration. Using three reference points S 1 , S 2 , and S 3 with known locations and distances d 1 , d 2 , and d 3 to the target object, the object can be located at the intersecting point of the three circles. However, in a dynamic system where distance measurements are noisy and fluctuate, the task of localizing becomes difficult. This can be seen in Figure 1B, where with fluctuating distances, regions within the circles become possible locations for the tracked object. In this case, rather than the object being located at a single point at the intersection of the circles as in Figure 1A, the object can be located anywhere in the dark overlapped region in Figure 1B.

Power minimization is a serious issue in wireless sensor networks to extend the lifetime and minimize costs. However, in MotivationsWireless sensor networks are becoming increasingly prevalent in a wide range of areas from surveillance [6] to monitoring temperature, humidity, and other environmental parameters [12,10].Wireless sensor network are usually comprised of nodes powered by batteries in locations where maintenance access may be difficult. Minimizing energy consumption in these networks would go a long ways toward extending the lifetime of the network and increasing the usability. However, in order to obtain a thorough understanding of the energy consumption characteristics in these networks, accurate modeling methods need to be developed.Although the primary energy consumption in wireless sensor networks is for communication [2], the processor also serves a key role and there is a need to examine the energy characteristics of the embedded processors of these sensor nodes. A sound method of modeling provides a stable platform upon which the energy characteristics of innovative technology can be analyzed. One example of this is the capability of processors to power down to a low power mode after some time of no activity, which we will refer to as the Power Down Threshold. However, once powered down, the processor requires time Power Up Delay to reach operating mode again.In this paper, we compare two different processor models: a Markov chain and a Petri net against software simulation. We show that the Petri net model is more accurate than the Markov chain. Section 2 introduces the use of both Markov models and Petri nets. Section 3 presents related work. Section 4 develops a model of a processor using a Markov model and a Petri net. Section 5 compares the results obtained by the simulation, Markov model, and Petri net, and Section 6 concludes this paper. Introduction to the use of Markov Models, Petri Nets, and Software SimulationMarkov models have long been used to model systems dealing with exponentially distributed arrival and service rates. Markov models are composed of event chains where the likelihood of a system being in a particular state is independent of being in any other state. We can see how this restriction limits the usefulness of this method. Petri nets, on the other hand, are far more useful than Markov models. Initially Petri nets were based on Markov models, but since then, features of Petri nets' have been extended over the years that has blurred these similarities. Unlike Markov models, Petri nets are a simulation based approach. Petri nets can be thought of as a directed graph of nodes and arcs which are called places and transitions respectively. With Markov models, closed form equations can be derived that can be used to provide analysis about the modeled systems. Petri nets require that the modeled system be simulated for extended periods of time so that the steady state probability of the system is reached. Computer program such as TimeNet 4.0 [13] are available that...

This paper presents a survey of energy efficiency of Medium Access Control (MAC) protocols for Wireless Body Area Sensor Networks (WBASNs). We highlight the features of MAC protocols along with their advantages and limitations in context of WBASNs. Comparison of Low Power Listening (LPL), ScheduledContention and Time Division Multiple Access (TDMA) is also elaborated. MAC protocols with respect to different approaches and techniques which are used for energy minimization, traffic control mechanisms for collision avoidance are discussed.We also present a survey of path loss models for In-body, On-body and Off-body communications in WBASNs and analytically discuss that path loss is maximum in In-body communication because of low energy levels to take care of tissues and organs located inside the body. Survey of Power model for WBANs of CSMA/CA and beacon mode is also presented.