WP 4: Next-generation sensing and fusion platform, Dr. Ketan Rajawat
Motivation and context:
The wireless sensor network constitutes an integral block of the proposed monitoring framework. The past decade has witnessed a plethora of WSN deployments for a number of water monitoring applications. General purpose WSN hardware, middleware, and software are now readily available for use in such monitoring platforms. However, large-scale and very large-scale deployments continue to be challenging, and have thus far only been implemented for specific applications, such as ocean monitoring.
This work-package aims at the development of the next-generation sensing and fusion platform for wide area river monitoring. Different from the state-of-the-art WSN deployments, the proposed system will allow hands-free, real-time monitoring, adaptive sampling, scalable deployment, and bad-data detection and elimination paradigms. The collected data processing will enable construction of physio-chemical maps, enabling identification, monitoring, and control of contaminant sources and flood management.
The salient aspects of the proposed system desigo are as follows:
1. Protocols for low-power sensor operation: the proposed WSNs will feature novel MAC layer protocols that will enable low-power operation at the sensors. Different from the traditional approaches that employ fixed duty cycling strategies for conserving energy, the proposed protocols will perform adaptive and dynamic duty cyclic adjustments that will minimize the energy consumption while keeping the estimation accuracy at a pre-specified threshold.
2. Adaptive sampling via feedback: The WSN platform will be equipped with the capability to incorporate feedback from the control center. Manual feedback can be provided by the monitoring authority at the control center in response to critical parameter readings. For instance, sampling with higher temporal resolution can provide high-accuracy estimates for certain parameters, which may subsequently be used for making control or policy decisions. Likewise, automatic adaptive sampling may be used to ensure that the estimation accuracy stays at a pre-specified threshold, especially for seasonal variables such as water level and water quality.
3. Compression for low-power operation: We will investigate the use of an additional compression step for reducing the power requirements at the sensors. The optimal compression rate will be inferred from the calibration data of the sensors, and will be used to compress the data prior to transmission. To this end, the trade-off between the computational effort required for compression and the saved transmit power will be studied. Both lossless and lossy compression mechanisms will be investigated.
4. Optimal relay placement: Communication relays are often required at locations with little or no network coverage. As part of the project, the problem of energy-efficient relay placement will
also be investigated. The idea here is to optimally place the relay nodes, while maximizing the sensor node’s battery replacement time.
5. Two-way communication paradigm: The proposed WSN platform will allow two way communication between the CC and the sensing nodes. On the one hand, the sensors will transmit parameter readings and diagnostic data to the CC. On the other hand, the sensors will be equipped to receive configuration data from the control center. It will therefore be possible to remotely configure the sensors to operate at various configurations (e.g.: sampling times), run diagnostic tests and send specific control commands to the nodes in real time. Such commands can be sent manually or in an automated manner.
6. Data collection and display: The entire data will be collected into a central database and will be available through a portal in near-real time. The users will be able to view various plots pertaining to the data, as well as access historical data over any period.
7. Data Analytics: As a final post processing step, the portal will feature the capability to create spatio-temporal maps of various parameters in online and offline modes. To this end, we will make use of state-of-the-art filtering and interpolation techniques to generate such maps with high spatio-temporal resolution. The proposed tools will automatically handle missing data points, carry out data de-noising, and filter bad data and anomalous measurements. The developed data analytics and learning tools will not only ensure succinct visualization of various physio-chemical parameters, but also allow control actions to be taken in a timely manner by creating policy-based alarms.
1. Distributed low-power WSN protocol capable of two-way communications
2. Algorithms for compressive and adaptive sampling
3. Relay placement design algorithms
4. Portal for data collection, visualization, and analytics