A wireless sensor network (WSN) is a collection of sensors which sense their environment, collect information and communicate this information to each other or to a decision center wirelessly. WSNs play important roles in a wide range of applications including environmental monitoring, surveillance, process monitoring, body area networks and smart homes. A growing concern in today's sensing systems is the efficient usage of energy resources. In that respect, solutions that consider energy harvesting (EH) offer a promising perspective. Sensors with EH capabilities use the other available energy sources, such as solar power or mechanical vibrations, instead of completely relying on a fixed battery or the power from the grid.
EH capabilities bring new engineering challenges for the design of sensing and communication strategies. When the energy to be harvested comes from renewable sources, (predictable yet still) unreliable characteristics of these sources is the key issue. Recently, there has been a significant effort to understand the information transfer capabilities of sensing and communications systems with EH capabilities. Here the mainstream approach is to focus on the difficulties regarding the reliable communications. These works typically consider an information-theoretic framework with the rate maximization as the performance criterion. This line of work is important for understanding the fundamental limits regarding communications with EH systems, yet it cannot adequately exploit the interactions between sensing and communications. An example of this interaction is the trade-off between how often the measurements should be made and how often they can be sent. This project will address these issues by considering sensing and communications in a joint framework. The main aim of the project is to provide guidelines about optimal sensing and communication strategies for systems relying on EH sensors. We will consider a unified framework that jointly optimizes sensing and communication strategies. The questions of ``when and how to make measurements, i.e. sense'' and ``when and how to send, i.e. communicate'' will be treated jointly. The underlying performance criteria is accurate reconstruction of the unknown field, i.e. estimating the unknown values of interest, such as humidity over an entire farm area, as accurately as possible. For data transmission, low-complexity signal processing techniques (linear precoding, power allocation) will be emphasized. Delay constraints, energy costs of sensing, battery size constraints and battery imperfections will be important ingredients of our framework. We will answer questions such as ``for a given energy reliability of EH source, when, where, and with what accuracy the samples should be taken?'', ``how reliable should the energy harvesting sources be to achieve a target distortion for field recovery?''.
We will first consider the scenarios where the energy that can be provided by the renewable energy sources can be predicted accurately. Then we will move onto more challenging scenarios where only statistical knowledge about the energy sources is available. We will also do measurements to characterize the energy availability from renewable sources for low power devices. This data will be used to evaluate our assumptions about energy arrival processes and the performance of our proposed solutions. With its unified framework that considers sensing and communication aspects jointly, its emphasis on low-complexity solutions and the complementary measurements for characterizing the energy availability, our project will provide us comprehensive engineering insights about energy harvesting systems that cannot be provided by the current approaches. These results will give us guidelines for our decisions about when and how to include renewable energy sources in our future sensing systems.
Professor at Signals and Systems, Signal Processing
Forskare at Signals and Systems, Signal Processing
Funding years 2016–2019
Area of Advance
Chalmers Driving Force