IoT-Enabled Water Distribution Systems—A Comparative Technological Review

Water distribution systems are one of the critical infrastructures and major assets of the water utility in a nation. The infrastructure of the distribution systems consists of resources, treatment plants, reservoirs, distribution lines, and consumers. A sustainable water distribution network management has to take care of accessibility, quality, quantity, and reliability of water. As water is becoming a depleting resource for the coming decades, the regulation and accounting of water in terms of the above four parameters is a critical task. There have been many efforts towards the establishment of a monitoring and controlling framework, capable of automating various stages of the water distribution processes. The current trending technologies such as Information and Communication Technology (ICT), Internet of Things (IoT), and Artificial Intelligence (AI) have the potential to track this spatially varying network to collect, process, and analyze the water distribution network attributes and events. In this work, we investigate the role and scope of the IoT technologies in different stages of the water distribution systems. Our survey covers the state-of-the-art monitoring and control systems for the water distribution networks, and the status of IoT architectures for water distribution networks. We explore the existing water distribution systems, providing the necessary background information on the current status. This work also presents an IoT Architecture for Intelligent Water Networks - IoTA4IWNet, for real-time monitoring and control of water distribution networks. We believe that, these components need to be designed and implemented effectively to build a robust water distribution network.

total amount of extracted water. Water leakage is primarily 94 caused by the aging of the pipeline (corroded), excavation 95 across the road, high pressure across the pipeline, unnoticed 96 underground pipe leakage, extreme weather conditions, mate-97 rial of construction, soil conditions, increased water con-98 sumption of systems that are at capacity and/or are already 99 stressed, increased treatment costs due to larger treatment 100 system requirements, capital expenditure, chemical require-101 ments, operational expenditure, and increased power require-102 ments through larger capacity treatments and pumping assets 103 to support the higher demand [16]. To reduce the leakage, 104 a proper monitoring and maintenance of the water distribution 105 system is required. Moreover lack of historic scientific data 106 on distribution system can leads to improper management and 107 maintenance of the resources and consumer connections such 108 as supply deficiency, leakage, higher demand, low pressure. 109 Table 2 depicts various stages of the WDS and its parameters 110 to be monitored in each stage to achieve better functioning 111 and equitable supply. Changes in public priorities, emerging 112 technologies, energy costs and increasing complexity are also 113 challenges for the WDS. In this paper, we are interchangeably 114 using the terms water distribution system and water supply 115 systems. preferably via any network (route) and using any service [29].
Introducing IoT into WDN can tackles one of the 167 critical challenges in water management; data unavailabil-168 ity/inadequate scientific information on various WDN ele-169 ments such as water reservoirs and network health. The real 170 time data collected by heterogeneous interconnected devices 171 and sensors can analyze and process the application environ-172 ment information, hence IoT would be a promising solution 173 for the WDN. Moreover, the application of IoT technology 174 can prove valuable in better water collection, storage, distri-175 bution, leakage prevention, waste water management, as well 176 as distribution [44]. 177 Table 3 summarizes the existing review articles on the 192 methodologies for monitoring technologies in water distri-193 bution systems. They provide different orientations, such as 194 these two articles [1], [49] focus on water quality monitoring 195 systems based on WSN and IoT systems, information and 196 communications technology (ICT) in water supply systems 197 in terms of the importance of the network pressure man-198 agement and stakeholders engagement via social media and 199 gaming [50], [51], and blockchain solutions [52]. However, 200 most of the review articles do not consider IoT solutions 201 for water distribution systems with monitoring, control, and 202 automation. 203 D. CONTRIBUTIONS 204 The management of water distribution is critical for the utili- 205 ties and authorities, as they need to get a thorough analysis of 206 the distribution network with the scope of IoT to embrace the 207 system flawlessly. In this paper, we have explained the water 208 supply/distribution system with the pervasive inclusion of the 209 IoT and presented the applications. This survey endeavors 210 to examine the research carried out to provide a sustainable 211 water distribution system by utilizing the features of IoT. The 212 major contributions of this paper are as follows. 213 • Presentation of the water supply system taxonomy, its 214 components and parameters, and their different pro- 215 cesses at each stage. 216 • Providing extensive insights into the IoT communication 217 technologies, cloud platforms, and their characteristics 218 applicable to water distribution systems. 219  TABLE 3. Summary of some popular survey articles in the area of IoT and water distribution systems. knowledge, we have considered all the above aspects 240 together, which provide a holistic approach that turns out to 241 be novel in this review paper. We feel that this paper will 242 help newcomers to understand the role of IoT in developing 243 a sustainable WDS. 244 E. ORGANIZATION 245 The organization of the paper is as follows. Section II briefly 246 explains the research method used in the study and develop-247 ment of the research questions such as resources for the study, 248 thought process for the research development, and finally 249 presents the research questions. Each research question is 250 addressed and discussed in detail in Section III. Section IV 251 comprises the analysis of IoT characteristics and constraints 252 in the deployments of WDN. Section V explains the ser-253 vices and techniques to facilitates WDN by presenting IoT 254 enabled framework for WDN. Section VI recommends an 255 IoT architecture capable of monitoring, control and automate 256 WDN systems and explains the existing challenges and future 257 direction. Section VII concludes the paper.

259
In this section, we introduce the research approach, by con-260 ducting a systematic literature review to understand the role 261

317
In this section, we will discuss the above-mentioned four 318 research questions and present the findings of each. The water supply system is a spatially organized network that 323 ensures safe water access to the people/community, which 324 consists of water resources, intake system, storage system, 325 conveyance systems, treatment plants, distribution networks, 326 and consumers. The water intake from the source is followed 327 by the transmission system in which conveyance of the raw 328 water is carried out from the collection unit to the water 329 treatment plants (WTP). Subsequently, the distribution of the 330 treated water to the consumers through the pipe networks. 331 In a further attempt to provide more insights into the water 332 distribution system, we portrayed the water supply system 333 with stages, processes, and its attributes and parameters in 334 Fig. 2. The water supply system consists of the following 335 seven subsystems.

337
The commonly used water sources are rivers, lakes, aquifers, 338 and bore wells. These water sources can be of different 339 types such as surface water sources, groundwater sources, 340 FIGURE 2. Water supply system taxonomy, consisting of different processes in the supply system from the water resources to the stakeholder. VOLUME 10, 2022 atmospheric water, recycled wastewater, and saline water.
presence in the intake tower (wet and dry intake) and the 355 position of intakes (submerged and exposed intake) [62].

394
• Coagulation and flocculation: Respective steps intended 395 to overcome the forces stabilizing the fine suspended or 396 colloidal particles, allowing particle collision and floc 397 development.

398
• Filtration: Water is passed through sand or multimedia 399 filters for the removal of left-over suspended solids and 400 micro-flocs.

401
• Disinfection: Disinfection is the process to remove, 402 deactivate or kill pathogenic microorganisms so that 403 they are not infectious to humans and animals. The water abstracted from the source may not be of usable 406 quality in its natural state. Moreover, anthropogenic activities 407 in many regions and the industrial, agriculture, and social pol-408 lution [64] compel water quality deterioration. Thus, the qual-409 ity of the supplied water should ensure public health safety 410 (essentially free of disease-causing microbes and chemicals). 411 Some of the widely used water quality guidelines [65]  Since IoT is a highly advanced technology, it can trig-503 ger the development of intelligent devices, smart sensors, 504 actuators, and M2M devices [75], [76] with the coexistence 505 of different IoT technologies like Near Field Communica-506 tion (NFC) [77], ZigBee [78], Wi-Fi [79], [80], LoRa [81], 507 and NB-IoT [82]. The IoT communication technologies are 508 intend to connect heterogeneous objects or devices within one 509 framework to achieve smart applications and services, with 510 low cost and low power even in adverse communication envi-511 ronments such as in lossy and noisy communication links. 512 To provide more insights into the different IoT technologies 513 and their characteristics, a comparison is shown in Table 5. 514 From different IoT communication technologies, RFID is 515 considered as the first IoT communication technology, which 516 realized the M2M communications via RFID tag and reader 517 concept [83]. NFC is another technology that supports high 518 frequency, low data rate communication with an applicable 519 range up to 10 centimeters [84] which allows the shortest 520 communication distance. The Bluetooth technology is known 521 as the cable replacement technology which is widely used 522 in headphones, mouse, and keyboards while Bluetooth Low 523 Energy (BLE) is used in accessories for smartphone apps and 524 internet-connected devices [ Besides the communication technologies, other important 531 aspect of IoT is the cloud platforms and protocols. There are 532 a number of public and private cloud platforms available for 533 IoT applications. Most of the cloud platforms are providing 534 end to end connectivity and services to the edge network 535 and end devices. The major focuses of the cloud platforms 536 are connectivity and normalization, device management, pro-537 cessing and decision making, data visualization and analytic, 538 external interface and database management [95]. The cloud 539 platforms are supported by various data transfer protocols 540 such as Message Queuing Telemetry Transport (MQTT) 541 that handles publish subscribe for message broker, Hyper-542 text Transfer Protocol (HTTP) and so on. Prominent cloud 543 platforms and their supporting protocols are shown in the 544 Table 6. 545 The IoT system has different functional blocks like iden-546 tification, sensing, services [48], management [96], secu-547 rity [97], semantics [38], and applications. Moreover, each 548 IoT system needs to accomplish the major IoT characteristics 549 like interoperability, scalability, QoS, reliability, distributive, 550 and security [98]. Furthermore, the steps involved in the 551 IoT system development are understanding the necessities 552 and requirements of IoT users (consumer/utility/stakeholder) 553 and their appliances and devices, pervasive communication 554 networks, and software architectures to transmit, process 555 and compute the sensor data to where it is relevant, and 556       The primary objectives of IoT, which will helps to con-692 struct a proper monitoring system, sensor and device integra- IoT-based water distribution system operation are as follows.

696
The initial step is to facilitate a visualization schematically 697 to gather sufficient information of water supply network 698 components such as, pipes, pumps, air valves, tanks, and 699 stabilizers, to group them in the next step in the geographic 700 information framework [141]. Next, to continuously monitor 701 the various water supply system parameters such as water 702 flow, pressure, and quality a set of sensors are deployed. 703 Finally, the sensed parameters are transmitted via communi-704 cation channels to an information system for the analysis and 705 to take suitable action [99]. Fig. 3 presents the closed-loop 706 strategy for an IoT network to attain the automatic control 707 features for the the system. An automated real-time control 708 systems can be achieved by the closed-loop strategy with the 709 feedback system. Further, the analysis of monitored results, 710 the hydraulic simulation strategies and the prediction of the 711 control variables according to the control factors can enhance 712 the efficiency of automated real-time WDS.

713
The hydraulic simulation and optimization strategies in 714 water distribution system, deals with demand prediction, 715 network design, pump operation, and real-time processes. 716 Although, major advances were made in this area, these 717 are still unexplored (or poorly explored) methodologies 718 that can be tested and applied in a considerable number 719 of water systems. Furthermore, AI models are power-720 ful tools in hydrology that can facilitate reliability, cost-721 effectiveness, problem-solving, decision-making, efficiency, 722 and effectiveness.  [159], (c) detection and limiting of the leak-731 age and cost-saving [160], (d) increase the business efficacy 732 via automating the traditional processes and functions [161], 733 [162] and, (e) the water quality improvement [163]. A major 734 challenge that the WDS has to conquer is to provide the 735 utility with the required information in a rapid manner. 736 Therefore AI possesses significant potential to address the 737 urgent challenges encountered by the WDN. Over previ-738 ous decades, there has been a considerable of research and 739 applications of IoT, including in (a) intelligent distribution 740 network [164], (b) robotics [165], [166], (c) WDN optimiza-741 tion management [167], [168], (d) automation [169], [170], 742 and (e) knowledge-based systems and decision support sys-743 tems [171], [172], [173]. The list of various WDN applica-744 tions, their monitoring parameters and used machine learn-745 ing (ML)/AI techniques listed in Table 10. Further, we list 746 the real-world water supply system monitoring platforms in 747 Table 9.     connectivity, data management, and security measures 768 across devices, actuators, and sensors cloud servers and 769 end-user interfaces in the system. This also reduces the 770 gaps between the protocols, hence improves the security 771 as well as reduce the overall cost of data [178]. • There exists compatibility between the water supply 786 infrastructure which is a legacy system to the specialized 787 devices, field equipment, and software, and so forth. 788 As data synchronization and data reliability are more 789 important, the scaling of the IoT networks and IoT 790 devices are critical [181]. Therefore a systematic design 791 and development of the software can standardize the 792 analysis of the generated data, code refining, and feature 793 introduction and upgrading [182].    sensing field variables using an IoT communication network. 849 The sensors (see Table 8) are to monitor the stimuli and 850 respond to events in the water supply systems. The sensors 851 or actuators can achieve two-way communications within 852 the network by providing commands to be sent from the 853 water utility to the smart sensors for various functionali-854 ties, including real-time monitoring of parameters such as 855 water flow at the pipes, pressure at the nodes and changing 856 the frequency sampling of readings. Furthermore, short-term 857 deployments of devices/sensors powered by batteries, and 858 long term deployments can be powered by solar panels due 859 to their low-power consumption characteristics in the sensing 860 layer. Moreover, the control part of the sensing layer can 861 acts as a data sink, transceiving data from the communica-862 tion layer. The data/information received to the control layer 863 can alter the actuators state. The communication network 864 between the sensors and the utility center intends to collect 865 and distribute the relevant information to consumers, suppli-866 ers, stakeholders, utility companies, and service providers. 867 Each sensor has a specific communication technology, which 868 is dependent on the climatic and geographic (spatial and 869 VOLUME 10, 2022   IoT gateway based either on Ethernet or other communi-889 cation technologies (see Table 5) Wi-Fi, WiMAX, Zigbee, 890 mobile communications, LoRa, RFID, and Bluetooth Low 891 Energy (BLE), NB-IoT, 5G etc [203], [204], [205]. This 892 layer has the field gateways which interfaces IoT gateways or 893 edge nodes with transceivers using ZigBee, Bluetooth, NFC, 894 Wi-Fi, LoRa, or Sigfox.

896
The service layer serves as the interface for both the 897 consumer and the IoT system. The service layer has two 898 sub-layers, IoT services sub-layer and analytical services 899 sub-layer. The IoT services sub-layer handles data ingestion 900 from the communication layer and the analytical services 901 sub-layer handles data processing (digestion) and perform 902 various analytics. The IoT services sub-layer provide ser-903 vices for the system to achieve device management [206], 904 data acquisition [207], device discovery [208], [209], remote 905 sim provisioning [210], [211], platform hosting [33], [212], 906 [213], and computer vision [214], [215]  such as water quality monitoring, treat-945 ment processes monitoring, the data analysis leads to the 946 control measures, and automation of the control process 947 (e.g., Chlorination). Environmental monitoring, water dis-948 tribution network infrastructure monitoring (asset monitor-949 ing), leakage detection, the anomaly detection for the entire 950 network by analyzing both spatial and temporal scale, con-951 sumption analysis, and demand prediction are the differ-952 ent application-level functionalities for the water distribution 953 system.

954
The components required for the implementation of Intel-955 ligent IoT based water distribution network shown in Fig. 5. 956 This encapsulates most of the studies analyzed in this paper 957 such as IoT sensors, communication technologies, protocols, 958 architectures, optimization techniques, data analysis methods 959 and types of middle-wares. With all these insights we are 960 recommending an architecture for intelligent IoT based water 961 network. 962 VOLUME 10, 2022  Table 10). Reservoir monitoring is required to determine 990 the quantity, quality, seasonal variations, and optimized water 991 allocation. Treatment plants are one of the essential entities in 992 WDN as the change in water quality can adversity affect the 993 health of consumers as well as WDN infrastructure. The qual-994 ity and quantity assessment of portable water, usage pattern 995 analysis, water demand profile maintenance and prediction, 996 the discovery of malpractices, and automated water meter 997 readings and billing are some of the services that come under 998 utility that are intended for the consumers. For enabling all 999 these services, WDN monitoring, automation, and control are 1000 essential.

1001
An example scenario for the IoTA4IWNet architecture is 1002 as follows; sensors are deployed to monitor water pressure 1003 within a distribution network. A single sensor out of a thou-1004 sand deployed, sensed the pressure variation in one single 1005 node due to a water hammer (single instance), thus raising 1006 a false alarm. The edge node closest to that sensor reacts 1007 immediate due to proximity, however, a hierarchically higher 1008 fog node at the city's observatory office collates all responses 1009 from sensors and passed it to the cloud. Thus, a predictive 1010 judgment is made based on machine intelligence. This pre-1011 diction can be used for future occurrences.  consists of devices that can enable connectivity between edge 1071 to fog, fog to cloud and within fog plane. The fog computing 1072 devices can get the context-aware computing paradigm by 1073 moving the intelligence to LAN level and data processing 1074 at fog plane. The communication interface layer enables the 1075 communication between fog nodes using LPWAN technolo-1076 gies and communication interface for fog plane to cloud plane 1077 using 4G/5G/6G technologies.

1078
IoT Fog Services: The services provided by the fog plane 1079 includes storage (mini data centers) and computation of the 1080 data. The data from the edge plane processed in fog plane and 1081 enables the services to both edge plane and fog plane. Various 1082 computing algorithms run at the fog plane and extract remote 1083 intelligence of the data by the analytics. The anatytic service 1084 is vital for extracting the insights for future updation of the 1085 system. ML and deep learning are some of the tools used 1086 for analytics. Furthermore it provides integration services 1087 and user interface services. The integration services allow 1088 dynamic management and future development of the fog 1089 plane. It also provides the communication interfaces services 1090 for cloud and edge plane. The security service provides a 1091 trust relationship between fog plane to the edge plane and 1092 cloud plane by guaranteeing the required network security, 1093 communication security and integrated computing modules 1094 security. Refer Table 12 for security threats and mitigation 1095 measures for fog plane. The IoT cloud plane is introduced to deal with massive 1098 data (Big data). It is the most powerful plane in terms of 1099 computation, efficacy, storage and other resources. The cloud 1100 plane consist of network of data centers which accumulates 1101 with various application data. From the survey conducted (see 1102  Table 11) COAP, MQTT, AMQP and HTTPS are the com-1103 munication and application protocols that provide services 1104 based on Representational State Transfer Application Pro-1105 gramming Interface (REST API) [235]. The communication 1106 interface layer for data centers and cloud platforms consist 1107 of 3G/4G/5G/6G [236] networks. The IoT cloud platforms 1108 widely used for various water system application are given 1109 in Table 6. Since the IoT cloud layer deals with Big data, 1110 the analytic techniques can combine with the Big data, which 1111 can be structured, semi-structured, or unstructured. The data 1112 cleaning and autonomous data quality check are significant 1113 roadblocks to the WDN system due to the integration of 1114 several heterogeneous data sources/sensors. We can fine-tune 1115 the Big data analytics for WDN by including all the influ-1116 encing parameters, such as physical, chemical, biological, 1117 socio-economical, geospatial, and behavioral of the WDN 1118 system, to predict short-term to long-term changes. Data-1119 driven decision-making, scientific discovery, and process 1120 optimization of WDN can be achieved with the help of Big 1121 data analytics [237]. Demand forecasting, water leak pre-1122 dictions, reservoir capacity prediction, reservoir water level 1123 predictions, queries, reports, visualization and interpretation, 1124 modeling and prediction, service improvement, and auditing ities [243], [244]. Hence the cyber security for the WDN 1172 system should be ensured for the reliable water distribution 1173 networks [245], [246], [247]. The types of attacks on cloud 1174 plane and its existing measures for maintaining cyber security 1175 in the WDN-enabled IoT system are present in Table 12.   Table 13.

1184
A sustainable management of water distribution network 1185 requires both technical and hydrological dependent system. 1186 A water distribution network varies spatially and temporally, 1187 i.e, according to seasonal changes, the water availability 1188 varies in the case of water network. The water distribution 1189 network is a complex infrastructure with inter-dependent 1190 entities and parameters, hence challenging the IoT integration 1191 on top of the network. The non-invasive sensors and control 1192 modules preferred over invasive modules as health of the 1193 infrastructure is delicate. Most of the real time implementa-1194 tions are pilot set ups and lacking end-to-end real time system. 1195 The scalability, security, implementation cost optimization, 1196 maximum bandwidth utilization via virtualization are the 1197 future directions of this application.

1198
The real-time monitoring and control of the WDN is chal-1199 lenging due to dynamic spatio-temporal variabilities. Health 1200 monitoring of each entities in the network such as pipe, 1201 valves, pump, nodes, and so on, due to cost effective and 1202 lack of appropriate IoT compatible sensors and devices. This 1203 can be rectified by design and implementation of a full 1204 fledged monitoring system, in a framework of optimized 1205 available sensor usage and incorporated software advance-1206 ment (AI based ML algorithms). Real-time demand moni-1207 toring, forecast and the implementation of the algorithms are 1208 primitive in WDS area.

1209
Several factors include seasonal variations, climatic 1210 changes, types of geographical regions such as urban, 1211 VOLUME 10, 2022 higher in the summer than in other seasons [262]. However, 1216 in some nations like Netherlands, water usage is higher in 1217 winter as the heater always works [263]. Likewise, climatic 1218 changes influence water demand variations. For example, the 1219 water requirements on a rainy day are lesser than on a humid 1220 day. Moreover, the daily water demand varies for rural, and     important to analyse the impact of risk assessment due to the 1268 pip burst or leakage. 1269 Water system starts with reservoir, hence it is important 1270 to monitor the quantity, quality, and health of the reser-1271 voirs. The system able to handle the dynamic characteristics 1272 of the reservoirs such as spatio-temporal variations climate 1273 changes (seasonal variability) and pollution due to waste 1274 dumping, industries, agriculture etc. A cost effective sustain-1275 able IoT-integrated intake system, which caters all the above 1276 parameters can mitigate the challenges. 1277 Water treatment plant monitoring, control and automation 1278 is one of the critical and complex task. The input water for the 1279 treatment plant depends on seasonal variability and the IoT 1280 system has to capture its dynamically varying quality param-1281 eters and also the systematic controlling and automation for 1282 each treatment process. Since the WDS is an spatial-temporal 1283 distributed system, the dynamic nature of the infrastructure 1284 makes the system unreliable for the static strategic function-1285 alities.
1286 Table 14 listed the challenges and recommendations 1287 required for IoT system design. Advantages of the IoT in 1288 WDN are enhanced transparency for the entire system pro-1289 cesses, immense response for identifying and predicting the 1290 anomalies and damages, optimized use of human resources, 1291 optimized cost, and sustainability (reducing water wastage, 1292 pollution and carbon footprint).

1294
This paper presented a detailed, up-to-date review of the state-1295 of-the-art of the role of IoT in water distribution systems. 1296 It presented the taxonomy of the water distribution system 1297 and role of IoT technologies, architectures, cloud platforms 1298 in various water distribution applications. It proposed an IoT 1299 architecture for intelligent water networks -IoTA4IWNet, 1300 for real-time monitoring, control and automation of different 1301 water distribution system applications. This work was guided 1302 by an exhaustive literature review and based on that, the 1303 research questions pertaining to IoT in water distribution 1304 networks and applications were formulated and analysed in 1305 great detail. During this study, it was discovered that most of 1306 the research focused on monitoring applications and did not 1307 cover closed-loop control strategy and prediction for water 5G-enabled IoT for industrial automation: A systematic review, solu-  Forum, Kyoto, Japan. In the last couple of years, from 2020 to 2021, Stanford 2241 University, USA, has ranked Dr. Maneesha as one among the global top 2242 2% scientists in the networking and telecommunication field for the quality 2243 of her research. Her work in multi-disciplinary areas such as the Internet 2244 of Things, wireless networks, wireless sensor networks, ML and AI, and 2245 sustainable development, successfully applied on thematic areas such as 2246 disaster management, water sustainability, healthcare applications, energy 2247 sustainability, and education access, has provided direct contribution for 2248 societal development.