On Maximizing the Data Volume in a Wireless Sensor Network with Time-Varying Channels

Abstract

Listen

Translate article

In this paper we present an analytical solution to the problem of jointly optimizing transmit power and scheduling for a star network of wireless energy-limited sensors around the base station. We have included factors which are usually neglected in modeling sensors and their communication channel. These include: accounting for power utilized for activities other than transmission, i.e. processing power, and time variations of the channel gains throughout the lifetime of the sensor. With an information theoretic approach to this problem, we have avoided the details of modulation and pulse shaping, and accounted for the mutual interference in simultaneous sensor transmissions. We employ the volume of data received at the base station over the lifetime of all sensors as the system performance measure and prove that in the optimal scheduling scheme sensors should avoid simultaneous transmissions. In addition, we show the optimal transmit powers that maximize the data volume can be determined via water-filling over the sensor's lifetime once the processing power and the single sensor transmission schedule are accounted for. In this work we have assumed prior knowledge of the channel gains throughout the lifetime of all sensors. Nevertheless, the solution will provide an optimistic value for the network's maximum data volume achievable by a causal scheduler. I. INTRODUCTION In gathering data from the coverage area of the network and conveying it to the base station, a wireless sensor net- work faces two main issues: mutual interference in sensor transmissions and limited battery energy. The problem of interference is particularly pronounced in the area surrounding the base station, because of the congestion produced by traffic concentration. Increased interference and congestion in this area, accelerates battery depletion in the sensors directly con- nected to the base station and reduces the volume of data that these sensors can convey to the base. Since their efficiency in communication limits the overall performance of the network, it is critical that we devise an optimal transmission scheme for these sensors. To improve our understanding of the factors involved and to eliminate unrelated elements, we segregate this area from the rest of the network and study it in isolation. In modeling sensors, aside from accounting for their limited energy source, we have taken into account the power utilized for functions other than transmission and termed it processing power. Although processing power introduces complications in the solution, it is an essential detail in employing a practical sensor model. To our knowledge this work is one of few that includes processing power in the sensor model.