Experimental Analysis of Cyber Physical System Perspective for Water Monitoring and Distribution

Abstract

Water plays an imperative role in proper functioning of the Earth’s ecosystems, and basically all individual activities, such as cultivation, manufacturing, transportation, and energy production. Abdul Kalam quotes that Water will become a very coveted treasure, more precious than gold and diamonds by 2050. In todays scenario pure potable water resources are depleting quite fast and human beings are ready to go to war to acquire hold on this natural resource. Hence, safeguarding(conservation) of this natural resources is given prime importance in my research work. The main cause of this is the rate at which human population is growing. If the rate of a growing population is compared with that of the potable water available on the earth, it logically concludes that we have to protect and preserve the precious water resource. Water conservation refers to the control, protection and development of the water resources both on ground and surface in order to prevent pollution. As water is the primary resource for life sustainability, it’s our responsibility to promote water management by conserving and using water efficiently. Additionally, it is our responsibility to save the water for our next generation and educate them to keep the water clean and safe for their upcoming usage. Physical water monitoring and management is a tedious and time consuming job. Alternately water management can be done using the available technology. Remarkable advancements in technology have so far addressed the water sustainability challenges using valuable tools. Key technologies include sensing technology, wireless communications and networking, hydrodynamic modeling, data analysis and control enabled intelligently wireless networked water cyber-physical systems (CPS) with embedded sensors, processors, and actuators that can sense and interact with the water environment. Unmanned water surveillance is possible with the help of communication technology. Real-time surveillance problem require intelligent organizations, planning, infrastructure design, water supply system and management. Smart water management is a good instance of real time CPS. For realizing this instantiation, calls a systematic framework together iiiwith the actual implementation of the associated modules is to be devised. In Cyber physical system there are 4 general working blocks: Monitoring unit, Networking unit, Computation unit and Actuation unit. In monitoring unit we have presented a systematic approach of customizing the existing water pipeline infrastructure using wireless technology. This is done to explore the possibility of wireless monitoring of the water distribution through a storage tank and later bring out the aspects of Cyber Physical system (CPS). To monitor the amount of water flows in storage tanks, we have carried out experiment at NITK, Surathkal, Mangalore, India. A set up of RF transceiver modules interfaced with water meter has been installed to get the water inflow and outflow details for each pipe. With respect to sampling time in the monitoring unit, a set up of another RF transceiver interfaced with aggregator and DTU will receive the water inflow and outflow details. This is then displayed on the web server. The networking unit includes the operation of software requirement and communication between data transfer unit (DTU) and aggregator unit (AU). The DTU and AU are present in the monitor unit to analyze the amount of water flowing in and out of storage tanks. The resultant data is available on a web server. The data aims to adapt ways and methods to overcome the improper handling and care of water flow in and out of storage tanks. In Computation unit, time series data of water flow rate has been used for forecasting. The water flow rate data is obtained at a sampling interval of 15 minutes and is classified into daily and monthly data. For time series data based water flow rate forecasting, we have considered and compared most prominent predictive models that are used in CPS. The forecasting models used are Simple Moving Average (SMA), Holt-Winters (HW), AutoRegressive Integrated Moving Average (ARIMA), Neural Networks (NN) and MultiLayer Perceptron (MLP). Performance metrics used are the root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The prediction obtained for daily data is observed to be more precise as compared to that of monthly data. In the case of instantaneous and cross-validation (10 fold), HW model and ARIMA models are observed to predict with low RMSE, low MAE and low ivMAPE, as compared to other models. In this work we have proposed a new hybrid model that is an ensemble of ARIMA and HW model. Simulation results indicate that the hybrid model is more accurate when compared with its basis time series model. This hybrid model can be used to conserve and distribute water when needed most. Real-time monitoring of water flow rates information helps us to conserve water when needed most hence we consider the first impediment in forecasting that is; handling missing data in the real-time monitoring system using the different imputation techniques. The performance is evaluated using traditional methods like RMSE, MAE and MAPE. Here two types of data missing are considered: 1) missing at random and 2) missing at burst. The validation is carried out by computing for a range of ‘data missing’, varying from 0% to 60% using different imputation techniques. This suggests that the proposed technique helps us to take necessary action to complete the decision-making activities (Computation unit) in CPS.