A Distributed, Energy-Autonomous Multi-Sensor IoT System for Monitoring and Reducing Water Losses in Distribution Networks

Abstract

Water resources are fundamental for human development in every possible sense; from natural development, since they are the main biological factor necessary for the development of life, to economic development, since they are essential for a large number of productive systems, especially in the primary and secondary sectors. This makes them a resource which, although at first glance may seem unlimited, is critical since their scarcity and unavailability compromise the whole of human development, greatly limiting productive and economic activity and, ultimately, social welfare. The current development of IoT technology, on the other hand, provides tools to face this problem in a technical way, allowing the adoption of distributed and automated solutions that, together with the knowledge provided by disciplines such as agricultural and alimentary engineering, make viable the development of a system that allows us to monitor and control water distribution networks (WDNs). Next, the situations that involve the mentioned problem will be detailed and different aspects will be proposed in which the implementation of the presented system is intended to have a direct impact.

1. Introduction

1.1. Background and Motivation

Water resources are fundamental for human development in every possible sense; from natural development, since they are the main biological factor necessary for the development of life, to economic development, since they are essential for a large number of productive systems, especially in the primary and secondary sectors. This makes them a resource which, although at first glance may seem unlimited, is critical since their scarcity and unavailability compromise the whole of human development, greatly limiting productive and economic activity and, ultimately, social welfare.

The current development of IoT technology, on the other hand, provides tools to face this problem in a technical way, allowing the adoption of distributed and automated solutions that, together with the knowledge provided by disciplines such as agricultural and alimentary engineering, make viable the development of a system that allows to monitor and control water distribution networks (WDNs). Next, the situations that involve the mentioned problem will be detailed and different aspects will be proposed in which the implementation of the presented system is intended to have a direct impact.

• Natural causes related to climate change: The decrease in precipitation during the past year, 2022, with respect to the reference data, between 1981 and 2010, reaches 16% on average among all the Spanish basins [1].
• Increase of water consumption: During the last years, Spain has experienced a notable growth in water consumption, both at the level of human consumption, which is reflected in the data provided by the National Institute of Statistics on household water consumption [2], as well as at the level of mass consumption in sectors such as agriculture, which represents 80.4% of the water consumed at the national level [3].
• Water distribution networks: As warned by the Drought Group of the Ministry of Ecological Transition of the Spanish government, the Spanish water distribution networks (WDN) are obsolete, producing a large amount of intangible losses, which implies a deterioration in the efficiency of the use of this resource [4].

In response to this context, this work proposes a distributed, energy-autonomous, multi-sensor IoT system designed to operate directly at the edge of water distribution networks. The solution integrates, within a single node, the combined monitoring of hydraulic variables (flow and pressure) and acoustic signals, long-range communication capabilities, and embedded hydraulic actuation.

Unlike existing approaches, the proposed system is not limited to passive supervision or centralized anomaly detection but instead extends monitoring and control capabilities to the pipeline level, enabling immediate and local actions on water flow. This architecture allows early responses to leaks, micro-leaks, or abnormal hydraulic behaviors, and establishes the technological basis for the effective mitigation of water losses, while reducing reliance on centralized SCADA infrastructures and delayed manual interventions.

1.2. Objective

The scientific goal of this work is to develop and validate a distributed IoT-based system capable of improving leak detection accuracy and enabling real-time monitoring and control in water distribution networks through the combined use of flow, pressure, and acoustic sensing together with autonomous valve actuation.

The aim is to design, build, and deploy a distributed embedded system capable of controlling and monitoring consumption in a water distribution network (WDN). It will have to be able to measure different physical variables related to the state of the waterway in which it is installed, such as the pressure and flow present, and it will also have to be able to control an opening and closing valve to determine the passage of water through the pipe. In order to detect leaks, the developed system will also implement a module for their detection through acoustic means, which will register the vibrations caused by the exit of pressurized water in the opening that forms the leak.

The developed system, together with the previously developed water quality monitoring system [5], will be an integrated solution that will allow us to delimit, study, and control the different existing problems related to the distribution and use of water.

Building upon this approach, the objective of this work is to develop and validate a distributed, energy-autonomous, multi-sensor IoT system that integrates hydraulic and acoustic monitoring, long-range communication, and embedded actuation within a single node. By extending monitoring and control capabilities to the pipeline level and enabling local decision-making, the proposed system is designed to provide early responses to leaks and abnormal hydraulic behaviors, thereby supporting the mitigation of water losses in distribution networks while reducing dependence on centralized SCADA-based control.

1.3. Main Contributions

The main innovative aspects of the beWater Drop system are as follows:

• Integrated multi-sensor monitoring at the network edge, combining flow, pressure and acoustic sensing within a single distributed node for leak and micro-leak detection.
• Embedded hydraulic actuation, enabling local, real-time control of valves and irrigation cycles instead of relying solely on centralized SCADA commands.
• Energy autonomy through solar power and battery management, which allows installation in off-grid or hard-to-access locations.
• Long-range LoRa communication and SCADA-compatible backend, facilitating scalable deployment of multiple beWater Drop devices over wide areas and seamless integration with existing water management infrastructures.

The presented system contributes solving the following problems, related to the mentioned situation of water distribution networks (WDNs), by the use of IoT technology providing data to its future analysis.

• Leaks and micro-leaks: Losses resulting from a deficient condition of the distribution infrastructure are responsible for the leakage of between 15% and 16% of the total water supplied, according to data provided by different media such as the Spanish National Institute of Statistics (INE) [2].
• Unauthorized consumptions: In situations of drought or low availability of water in the reservoirs, the regulatory government bodies usually impose restrictions on consumption quantities. In many cases, users consider the quotas granted to be insufficient and, in order to increase the productivity of their crops or not to pay for the water consumed, they carry out unauthorized prospecting on the distribution network. In fluid mechanics terms, an unauthorized consumption causes the same effect as a leak so that it will also be detected.
• Inefficient irrigation: Precision agriculture involves the efficient use of water in the irrigation process, calculating the precise amount of this fundamental resource and distributing it only in productive areas is essential to implement an optimized irrigation system. The implemented device is able to open and close its valve, controlling the amount of water released with precision, making use of an installed flow meter to calculate the amount of water in liters.

These contributions define the proposed system as a system-level invention that transcends isolated sensing approaches by integrating distributed multi-sensor monitoring and embedded hydraulic actuation at the network edge, enabling early intervention and supporting the mitigation of water losses in distribution networks.

1.4. Related Work

In this section a research about similar developments, both in academic and commercial areas.

The following review of academic and commercial solutions is structured to explicitly highlight the differences between existing approaches and the system proposed in this work. Particular attention is given to the level of integration between sensing, communication, energy autonomy, and control capabilities, as well as to the ability of each solution to go beyond anomaly detection and support the mitigation of water losses in distribution networks.

1.4.1. Academic Papers

• Water Management in Agriculture: A Survey on Current Challenges and Technological Solutions [6]: This survey, which investigates water management systems in agriculture. It sets out four challenges related to agricultural water management: water reuse and monitoring of water contamination levels, pipeline condition monitoring, irrigation management, and drinking water supply for livestock. Two of these four challenges are directly addressed: pipeline condition monitoring and irrigation management. This survey also establishes a reference deployment architecture, to which the proposed system is fully compliant.
• An Enhanced Water Pipeline Monitoring System in Remote Areas Using Flow Rate and Vibration Sensors [7]: In this article, a prototype with Wi-Fi and LTE communication capabilities is presented, which monitors, like the proposed system, the flow present in a water distribution system as well as its pressure and state through sensors that capture vibration. The main differences proposed with the system detailed in this article are the presence of the ability to control valves to open and close the way and the presence of a LoRa [8] communication module.
• Acoustic Leak Detection in Water Networks [9]: This article focuses on the analysis of spectrograms recorded by microphones installed on the surface of the pipes under analysis. It makes use of different analysis methods encompassed in the field of AI, such as convolutional encoders or adversarial networks. This is a test of the feasibility of analyzing the sounds recorded by a digital microphone using AI techniques, which is taken into account when designing the system. The inclusion of this microphone is therefore a justified decision based on studies that have been carried out previously.
• Leak detection using the pattern of sound signals in water supply systems [10]: This paper focuses on the analysis of audio signals recorded by a microphone arranged in the distribution network such as the one proposed in the present system. It obtains analysis data such as PCA, Principal Component Analysis, and DSF, Damage Sensitive Feature. These are used to perform leak detection using support vector machines. This paper is also a proof of the feasibility of using analysis techniques for leak detection based on vibration data captured by acoustic microphones installed in the distribution network.

To provide a clearer comparison of the main academic approaches described above,

Table 1. Comparative analysis of monitoring and leak detection systems for water distribution networks.

Reference	Application	Sensors Used	Detection Techniques	Communication Technologies	Energy/Autonomy	Architecture	Limitations	Differential Contribution
[6] Water Management in Agriculture (Survey)	Agricultural management (irrigation, reuse, quality)	Pressure, flow, acoustic microphones, water quality sensors	General supervision (no advanced techniques)	Wi-Fi, ZigBee, MQTT, etc.	Not addressed	Distributed architecture	Theoretical approach, no prototype	beWater Drop provides a real, autonomous prototype
[7] Enhanced Water Pipeline Monitoring	Pipelines in remote areas	Flow, vibration	Vibration analysis and flow comparison	Wi-Fi, LTE	Not autonomous	Basic prototype	Cellular dependency; no valve control	Adds solar autonomy, LoRa and hydraulic actuation
[9] Acoustic Leak Detection in Water Networks	Acoustic leak detection in urban networks	Contact microphones on hydrants	Acoustic detection and ML (autoencoders, adversarial nets)	Low-energy, low-bandwidth network	Not autonomous	centralized architecture. IoT nodes	High computational cost; no hydraulic integration	Lightweight acoustic analysis + hydraulic sensors
[11] AIoT-Driven Leak Detection	Urban real-world networks	Hydrophones + acoustic logger	Signal analysis + FFT	IoT, Wi-Fi with mobile network usage	Not autonomous	AIoT	Complexity and cost	Lower-cost, autonomous LoRa-based alternative
[10] Leak Detection Using Sound Signals	Water supply networks	Microphones	Acoustic detection	Laboratory-based approach	Not specified	Experimental	Limited to controlled environments	PCA/DSF feasible on low-power hardware
[12] Real-time Pipeline Monitoring with FSR Sensors	Urban multi-leak detection	FSR sensors	Multipoint analysis	Wireless IoT	Not specified	WSN	Limited scalability	Adds energy autonomy and hydraulic control
[13] Liu et al.—Time-Transformer for Acoustic Leak Detection	Distribution networks	Microphones	Sequential ML (Transformers)	Not specified	Not autonomous	Experimental-phase prototype	Algorithmic complexity	Prioritizes simplicity and low-power processing

summarizes their application domains, sensing technologies, detection techniques, communication capabilities, architecture, and limitations, together with the improvements introduced by the proposed beWater Drop system, and includes additional relevant references to offer a more comprehensive overview of the state of the art.

1.4.2. Commercial Devices

To contextualize the proposed system within the existing commercial landscape, this subsection first describes two representative commercial solutions—Plantae and Nautilus—which illustrate current approaches to distributed monitoring and leak detection. Building on these examples, the section then expands to a broader overview of additional commercial technologies available in the market.

• Plantae Flow Meter [14]: This device has a digital flow meter and a LoRa communications module. Unlike the proposed system, it does not have a pressure switch, a motor driver to control solenoid valves, or a microphone to detect leaks or other problems in the distribution network. It is able to measure, in a distributed way, the flow rates in different points of the distribution network and collect all these data in real time. If there are no leaks, the flow rate at the starting point must be equal to the sum of the flow rates in the different distribution paths. Through this simple set of sums, the system can check for the presence of leaks in real time.
• Nautilus [15]: This device is capable of detecting leaks in distribution networks of large magnitudes. Instead of being a device that is installed in the network to perform monitoring, it is a diagnostic device that travels along a pipeline and is able to detect the leaks present in its path. The device has a spherical shape that allows it to be introduced into the pipe to be diagnosed. Then, like the proposed system, it makes use of microphones to record the sounds present in the pipe. When the ball passes through a leak, peaks in sound intensity are recorded, which are stored together with the location of the ball. The main difference with respect to this system is that Nautilus is not able to warn when a leak occurs, since it is not permanently installed in the network and therefore cannot perform monitoring tasks.

Beyond the two representative devices discussed above, a diverse set of commercial products exists for leak detection and pipeline monitoring. To offer a structured view of this landscape,

Table 2. Comparative Overview of Commercial Devices for Water Network Monitoring and Leak Detection.

Device	Sensing	Communication	Real-Time	Autonomy	Valve	Install	Environment	Comparison
Plantae [14]	Flow	LoRaWan	Yes	Battery	No	Fixed	Irrigation	beWater Drop System adds pressure, acoustic sensing and valve control
Nautilus [15]	Acoustic	No communication. Creates offline datasets	No	One operation	No	Not installed	General WDN	beWater Drop remains installed, performing a WDN monitoring, while Nautilus performs a single scan when its deployed
HWM PermaNET+ [16]	Acoustic	Integrated cellular modem	Yes	Limited battery (5 years)	No	Fixed	Urban WDN	beWater Drop System offers multi-sensor fusion, long-range lora and hydraulic actuation
Flexim FLUXUS [17]	Flow	Local	Local only	Not specified	No	Clamp-on	Industrial Gas Distribution	beWater Drop System provides remote lora connectivity plus acoustic and pressure sensing
Echologics EchoShore-DX [18]	Acoustic	Cellular	Yes	Battery (5–10 years)	No	Fixed curb-box	Urban WDN	beWater Drop System integrates multi-sensor monitoring and distributed control
Gutermann ZONESCAN [19]	Acoustic	Nb-iot	Yes	Battery (5 years)	No	Fixed	Urban WDN	beWater Drop System improves autonomy and adds valve control capabilities
Sensus/Badger AMI [20]	Flow + pressure	NFC	Yes	Battery (20 years)	No	Fixed	Urban WDN	beWater Drop System adds acoustic sensing and active hydraulic actuation

synthesizes key characteristics across multiple commercial solutions, including sensing modalities, communication strategies, real-time monitoring capabilities, autonomous operation, installation requirements, and environmental targets. This comparative perspective highlights the gaps that remain in current technologies and underscores the contributions of the proposed beWater Drop system.

1.4.3. Relevant Patents

Patent literature highlights key innovations in leak detection and pipeline monitoring. For example, WO 2012/158267 A1 describes an acoustic probe for detecting leaks inside water pipelines using hydrophones and pressure sensors [21]. The European patent EP 3374747 B1 introduces a method for water leak detection based on pressure sensing and transient analysis [22]. Additionally, WO 2020/215161 A1 presents an apparatus for pipeline monitoring that integrates RFID-based sensing for temperature, humidity, and structural conditions [23].

Taken together, these patents demonstrate technological maturity in acoustic sensing, pressure monitoring, and wireless condition tracking. However, none of them combine, within a single distributed device, the integrated capabilities of flow sensing, pressure monitoring, acoustic detection, autonomous solar power, long-range LoRa communication, and embedded valve actuation, features that define the novelty of the proposed beWater Drop system.

2. Materials and Methods

2.1. Methodology

Methodology is used in the development process of this system. In the following table, the conveniences and inconveniences found that are related to this methodology are listed in

Table 3. Advantages and disadvantages found regarding the adoption of V methodology.

Advantages	Disadvantages
Standardization increases control over the development process.	It is a process with low flexibility, where significant deviations from the established plan are not allowed.
It increases reliability and precision when estimating development costs.	If any of the verification phases are not passed, it becomes a blocking factor for development, which can be severely affected in terms of deadlines.
The conclusion of each phase delivers a verified product that can be exported for outsourcing the development of subsequent phases.	It increases the product’s Time to Market, making it vulnerable to market disruptions and potentially jeopardizing its viability as a business project.
Codification is facilitated due to the existence of prior documentation and unit tests.	A larger team template is required because, ideally, the testing team cannot share personnel with the implementation team.
Tasks are appropriately assigned for each phase, providing a clear roadmap for development and minimizing idle times between phases.	Until the last phase, there is no prototype, so implementing changes to specifications once the product’s behavior is analyzed requires restarting development, thus consuming a significant amount of time.

.

2.2. System Architecture

In this work, we present beWater Drop, an IoT-based monitoring and control node for water distribution networks. Each beWater Drop device integrates flow, pressure, and acoustic sensing with autonomous solar power, embedded valve actuation, and long-range LoRa communication, extending conventional SCADA capabilities to the network edge. In the following section, we refer to the set of distributed nodes, gateway, and backend infrastructure as the beWater Drop system.

For carrying out the first phases of the V development model, SysMl [24] is used in requirement specification, system architecture design, and system behavior design. The resulting diagrams are available in the whole book summarized by this paper [25].

In Figure 1, SysMl’s block definition diagram is used to list the fundamental parts that compose the WDN monitoring and controlling system, where the present subsystem is responsible for controlling consumptions and detecting leaks. The WDN monitoring and control system is composed of different devices, such as the previously developed embedded system responsible for measuring water quality and detecting pollutants [5], a LoRa gateway responsible for collecting data provided by distributed devices, an MQTT broker, a time series database, and a graphing platform to display the collected data.

In Figure 2, the main components of beWater Drop system are listed. These are responsible for providing a solution to monitor and control WDN, detecting leaks and logging consumptions and other information related to WDN condition. First, the main components of the developed embedded system will be listed; these are responsible for collecting and processing data provided by the sensors and controlling the installed actuator. Then, a configuration server is listed. This server is supported by the ESP32-S3 and provides the user with a friendly interface to configure the operation of the embedded device. The water pipe device is composed by the different sensors and actuators that interacts with the WDN and provides data and control to the embedded device. Lastly, the power supply system makes possible to power up the system in electrical isolated environments. All of these components will be described in detail in the following section.

2.3. Hardware Development

The developed system is composed of several subsystems, which are described in detail in the different chapters of this paper. Each one of them has its own characteristics and, therefore, different criteria are used for its design. These characteristics are as follows:

• Embedded System: A PCB board has been designed and developed to interconnect the microcontroller, an ESP32-S3, with the rest of the hardware components of the system. In addition, the developed PCB implements in analog form some functions, such as the detection of sound intensity peaks in the signal captured by the membrane microphone installed in the system. The system responsible for making this detection is under the signal conditioner block in the previous BDD diagram and is composed by an amplification stage, a buffer, and several operational amplifiers as comparators in a Schmitt trigger configuration. This board is presented in the next Figure 3.

• Power supply system (Figure 4): The power supply system is composed of a lead acid battery, a BMS battery charge and status manager, and a solar panel that allows the system to be autonomous in environments that do not have a continuous 220 V AC power supply. Since this system is designed for field installation, the self-supply capability is crucial, as the need for a continuous power supply would significantly restrict the environments in which it could be installed. In the next figure, a photography of this system is provided. In this figure, there are two elements; firstly, a blue container box, designed specifically for this propose, containing the BMS, and the lead acid battery is on the back, while a solar panel stands in the front of the picture. Both the battery and the solar panel correspond to generic, commercially available components without a specific branded model, as they were selected for their compatibility with the system requirements rather than for any proprietary characteristics.

Figure 4. Power supply system composed by lead acid battery, BMS and solar panel inside of an ASA plastic 3D-printed container rugged box.

Figure 4. Power supply system composed by lead acid battery, BMS and solar panel inside of an ASA plastic 3D-printed container rugged box.

• Water pipe monitoring and controlling device (Figure 5): To enable the developed embedded system to control and sensor the water distribution medium in which it is installed, the presence of actuators and sensors on it are necessary, as they allow the physical interaction of the system. These are a digital flow meter, which senses the volume of water flowing through the pipe by generating pulses, a digital pressure sensor, capable of measuring the pressure at a specific time to which the pipe is being subjected, and a motorized ball valve, which is able to open and close, both fully and partially, the distribution channel. These are also accompanied by two microphones, one digital and the other analog, in charge of measuring the vibrations caused by possible leaks or micro-leaks present in the network. A sample device with only one microphone installed is presented in the next figure.

2.4. Software Development

This section will detail the different considerations carried out for the implementation of the different software components of the system. It will detail the different relevant aspects in the development of the embedded software running the monitoring and control embedded systems and gateway. It will also detail the implementations of the configuration portal, the server responsible for storing the data, the broker responsible for managing the connections with the different devices and the deployment control portal, where the different rules established to generate warnings and alarms are executed, which also allows monitoring of the data collected by the different deployed devices.

2.4.1. Embedded Software

In this section, the most relevant details about software development executed by the distributed embedded devices will be listed.

• Analog-to-digital conversion: The analog-to-digital converter ADC has to be used to measure the envelope signal of the one produced by the analog microphone, so it is necessary to use the highest sampling frequency that the ADC is able to give. In order to achieve this goal, firstly, a frequency of 3.4 MHz is set for the I2C line, corresponding to the High-Speed mode. Secondly, the interrupt reading is configured, which allows us to collect each sample at the precise moment it is captured. Finally, the ADC is configured to work at the highest number of samples per second (sps) that it offers, which is 3.3 K sps. This sampling frequency is not enough to fulfill the Nyquist theorem [26], so it is not useful to reconstruct the sound wave, but it is enough to perform a correct measurement of the envelope. For the pressure data measurements, the continuous reading by interruption is not used, the discrete mode with multisampling is used instead.
• Digital Microphone: To make use of the digital microphone it is necessary to correctly configure two elements; the I2S interface [27] and the direct access to memory (DMA) [28]. In addition, a file system capable of supporting the creation of .wav files with the captured waves must be configured. The I2S bus is configured at 16 kHz as sample rate, with a precision of 16 bits per sample, and the DMA buffer size is set according to the amount of data recorded. SPIFFS [29] is used to implement the file system responsible for saving the recorded .wav files within a custom partition table that allows us to save up to 25 MB. This table has also a partition reserved for allowing Over The Air (OTA) updates.
• LoRa transceiver: An RFM69 is installed as LoRa transceiver and is responsible for communicating acquired data to its configured gateway. For this proposal, different features are configured. First, a high power transmission is set, allowing us to expand the coverage of the connection; second, an ACK frame is added to the protocol, ensuring the reliability. Lastly, an AES [30] encryption is set, making the communications more secure.
• Configuration portal: A web server is implemented that supports a configuration portal programmed using HTML and CSS on port 80, since it is offered through the HTTP protocol. This portal allows us to change several settings on the operation of the embedded device, such as the periodicity of its measurements, recording time of the microphones, valve opening periods, and duration. This portal also offers the possibility to download the generated .wav files.

2.4.2. Servers

• Database: A TimescaleDB time series database is implemented due to its good performance [31], low cost, and integration with the other software components. This database is optimized to store telemetry data indexed based on the timestamp at which they are captured, thus optimizing their plotting and querying based on the time axis.
• Rule chain executor: For this project, two rule chains have been developed. The main one only executes the second one after saving the data in the time series database. The process that follows the second chain is recurrent and consists of executing a script that checks the value of the different variables, and depending on the result of the execution of this script, an alarm is created according to the problem that may occur, such as leakage, data out of the expected thresholds, etc. After this, if the alarm is created, a message is sent via email to the user to alert him.
• MQTT broker: This technology is used to carry out the messages collected by the gateway and upload them to the database. This protocol allows to define a topic structure for organizing the different deployed devices and its recorded data, making the message passing protocol scalable. For naming each topic, the number of the gateway is referenced, and the device number is then specified. So the devices “D” connected to Gateway “G” will publish in the topics below under the prefix GateWay<G>/Dp<D>/. After the prefix specified above, the different topics that will contain the information provided by the different sensors and actuators installed are defined.

2.4.3. Graphing Platform

One screenshot from the software implementation of the graphing platform through the Thingsboard(CE 3.x) tool is shown in Figure 6. In this example, the developed device is installed in an irrigation WDN. It shows the different irrigation cycles generated, in which the pressure measured by the pressure sensors, which typically capture a value of approximately 150 kPa, drops due to the depressurization of the pathway generated by the opening of the installed motorized ball valve. This graph also shows the flow rate present during each of the track openings, as well as two pressure gauges that show the pressure value at the time of the query.

In the upper left part, the identifier of the device from which the captured data is being displayed is shown, and the current status of the installed valve is shown on the right side.

2.4.4. Telecommunication System

For the development of the telecommunications system, Wi-Fi [32] and LoRa [8] communication technologies have been implemented, each of them being used for different purposes. In addition, the network topology used has been varied according to the requirements for each of the different application areas. In this chapter, the considerations that have been carried out will be detailed.

In Figure 7, the general architecture of the communications system can be seen in a simplified form, in which the use of the previously cited technologies can be appreciated.

First, within the field of deployment, corresponding to a water distribution system, it can be seen how an undetermined number “n” of distributed devices are installed, denoted under the name beWater Drop. On the right side of the diagram, a device is shown that connects to any of these beWater Drops through the ESP32-S3 installed in them. This connection, through an AP access point protected with WPA2 [32], causes the distributed device to raise an HTTP configuration portal, through which the technical team can configure the device. Secondly, it can be seen how these distributed devices in the deployment field are connected to a single gateway, which collects data from them. This communication is performed through the use of LoRa technology, which allows distances beyond 1 km radially around the gateway. In this case, the network topology used corresponds to a star scheme, with the gateway at the center of the network.

Once the data is available at the gateway, formerly called “Central” in the requirements’ specification, the data is downloaded via a point-to-point Wi-Fi connection to a server.

3. Results and Discussion

The experimental results presented in this section are discussed not only in terms of functional validation of the proposed system, but also with respect to their implications for the mitigation of water losses in real distribution networks. The combined use of flow, pressure, and acoustic data, together with embedded hydraulic actuation, enables early detection of anomalies and immediate local responses, which are key aspects for reducing reaction times and improving overall network efficiency.

Different tests have been carried out to ensure the validity of the developed system. Following the V methodology, unit tests, module tests, integration tests, and system tests have been executed to verify every step of the development. Finally, a test in a real scenario will be carried out to ensure that the system is capable of remote controlling and monitoring a drip irrigation water distribution line.

Beginning with unit tests, each of the used hardware elements is verified in an isolated and static environment; then, modular tests are carried out, where the different modules are tested with all its components working together.

3.1. Module Tests

In this stage the LoRa module is tested, composed by a transceiver and an antenna, the sound adquisition module, composed by two microphones, one analog and another one digital, an ADC and a signal conditioning stage, the flow monitor module, composed by the flow meter, pressure sensor, and the flow control module, composed by a motorized valve and a H-Bridge. All the module tests carried out are presented in the following table, and have been executed as dynamic white-box module tests.

In the next figure, results of the tests described in Table 4 are presented as images, showing the physical layout and obtained outputs.

Table 4. Planned modular tests set.

Module	Test Description	Test Results
ADC	Through the execution of a test program, voltage measurements are performed comparing the input against the output of a current-limited adjustable power supply. The values returned by the ADC must match the values provided by the power supply.	Passed. Verified at voltage levels: 1 V, 2 V, 3 V, 4 V, and 5 V. The power supply output matches the value captured by the test program and displayed via UART serial communication. (Figure 8a)
LoRa Transceiver	Point-to-point connectivity between the distributed device and the Gateway is validated via test software, ensuring data integrity by confirming that the transmitted sequence number matches the received value.	Passed. Field test results: Maximum coverage of 4 km achieved in an open, flat environment with minimal interference. Video available. (Figure 8b)
Digital Microphone	A test program is executed to verify that the digital microphone is capable of capturing an audio signal and generating a .wav file containing the received audio data. The file must be successfully stored in the file system.	Passed. INMP441 digital microphone setup and SPIFFS file system configuration verified. The recording was successfully generated and downloaded via the configuration portal. A spectrogram and the .wav file are provided as evidence. (Figure 8d)
Sound sensor	An analog microphone is connected to verify it generates a trigger when a sound of a specific intensity is detected. Additionally, the test verifies that the system’s sensitivity can be adjusted using the installed potentiometer.	Passed. Tested with surface impacts. Sensitivity was successfully tuned via the potentiometer. The module generates three signals: the amplified audio path with a 2.5 V offset (dark green), an intensity envelope proportional to volume (light green), and a threshold-based trigger (red) used to generate the system interrupt for sound analysis. (Figure 8c)
HTTP Server	Through the execution of a test program, verify that the HTTP portal is capable of accessing files stored in the file system and serving them in response to an HTTP GET request.	Passed. Validated in conjunction with Test 4. The output shows the portal address, the presence of the generated audio file, and its size, allowing for successful download. (Figure 8e)
Flow Control System	A test program executes control over the motor system. It must be verified that two motors—one connected to EV1 and the other to EV2—can be powered with 12 V in both polarities.	Passed. Multimeters connected to EV outputs verified voltage and polarity during sequential open/close actuation (12 V DC). (Figure 8f)

Table 4. Planned modular tests set.

Module	Test Description	Test Results
ADC	Through the execution of a test program, voltage measurements are performed comparing the input against the output of a current-limited adjustable power supply. The values returned by the ADC must match the values provided by the power supply.	Passed. Verified at voltage levels: 1 V, 2 V, 3 V, 4 V, and 5 V. The power supply output matches the value captured by the test program and displayed via UART serial communication. (Figure 8a)
LoRa Transceiver	Point-to-point connectivity between the distributed device and the Gateway is validated via test software, ensuring data integrity by confirming that the transmitted sequence number matches the received value.	Passed. Field test results: Maximum coverage of 4 km achieved in an open, flat environment with minimal interference. Video available. (Figure 8b)
Digital Microphone	A test program is executed to verify that the digital microphone is capable of capturing an audio signal and generating a .wav file containing the received audio data. The file must be successfully stored in the file system.	Passed. INMP441 digital microphone setup and SPIFFS file system configuration verified. The recording was successfully generated and downloaded via the configuration portal. A spectrogram and the .wav file are provided as evidence. (Figure 8d)
Sound sensor	An analog microphone is connected to verify it generates a trigger when a sound of a specific intensity is detected. Additionally, the test verifies that the system’s sensitivity can be adjusted using the installed potentiometer.	Passed. Tested with surface impacts. Sensitivity was successfully tuned via the potentiometer. The module generates three signals: the amplified audio path with a 2.5 V offset (dark green), an intensity envelope proportional to volume (light green), and a threshold-based trigger (red) used to generate the system interrupt for sound analysis. (Figure 8c)
HTTP Server	Through the execution of a test program, verify that the HTTP portal is capable of accessing files stored in the file system and serving them in response to an HTTP GET request.	Passed. Validated in conjunction with Test 4. The output shows the portal address, the presence of the generated audio file, and its size, allowing for successful download. (Figure 8e)
Flow Control System	A test program executes control over the motor system. It must be verified that two motors—one connected to EV1 and the other to EV2—can be powered with 12 V in both polarities.	Passed. Multimeters connected to EV outputs verified voltage and polarity during sequential open/close actuation (12 V DC). (Figure 8f)

Figure 8. Results of modular tests applied to different modules that are part of the tested system. Each subfigure is described separately in the table above, which provides the corresponding explanation for each element shown in the figure.

Figure 8. Results of modular tests applied to different modules that are part of the tested system. Each subfigure is described separately in the table above, which provides the corresponding explanation for each element shown in the figure.

3.2. Integration Tests

In the integration tests, all these different modules are tested together, with a single software flashed in the ESP32-S3. Once all the modules are working together, a system test is carried out for ensuring the full system functioning. In this test, the configuration portal is used for making manual openings and closings in the installed pipe. The graphing platform is also used in these tests for ensuring the validity of the data obtained by the sensors.

A full system operation log is shown in Figure 9, making an opening cycle, 3 min in duration, corresponding to a short irrigation. In this operation, pressure in the pipe must remain high while the solenoid valve is closed. Once it is opened, pressure data should be received and reduced in a linear fashion until it remains at a stable value. At this point, the flow meter should have generated a number n of pulses, which allows us to calculate the amount of water flowing through the pipe.

Once the irrigation time comes to an end, the solenoid valve must return to its normal closed status, and the pressure must reach levels similar to those prior to opening. All these behaviors can be seen in the figure, corresponding to the result shown in the graphing tool panel. In green, the opening status of the motorized ball valve can be seen, positive when it is open, negative when it is closed. In blue, the flow rate present in the lane during the opening period, and in red and orange, the values captured by the pressure sensors.

3.3. Real Scenario Test

Figure 10 shows an example of a real installation of the monitoring and control device within an operational drip-irrigation network. In this specific assembly, the components of the “Water pipe monitoring and controlling device” have been adapted to meet the requirements of the pre-existing irrigation infrastructure. Specifically, the standard motorized ball valve has been replaced by a solenoid valve designed to control water flow in 16 mm distribution lines, and the intermediate pipe sections were modified to match this diameter. Furthermore, traditional analog flow and pressure gauges have been installed to serve as reference systems for in situ verification.

In this case, the assembly is integrated into a distribution head that feeds a drip lateral supplying a blueberry plantation. The water flow follows the sequence described below. First, a manual shut-off valve is installed, allowing the water supply to be enabled or interrupted and facilitating maintenance operations or emergency interventions. Next, a digital flow meter is installed to measure the volume of water supplied and detect deviations from the normal operating flow range. Downstream of this point, the first pressure transducer is positioned to measure the inlet pressure upstream of the solenoid valve. The solenoid valve is then located downstream, responsible for fully opening or closing the water passage (with no option for partial flow regulation, unlike motorized valves used in other configurations).

A second pressure transducer is placed after the solenoid valve to record the outlet pressure, enabling verification of the valve’s state (open or closed) and allowing estimation of the head loss generated by comparing both pressure readings. The combined information from the flow meter and the pressure transducers enables the identification of hydraulic anomalies (such as increases in head loss or unexpected variations in flow rate). When these signals are correlated to the acoustic data captured by the microphone installed in the system, it becomes possible to detect leaks, micro-leaks, pipe bursts, or common issues in drip irrigation (such as emitter clogging or calcification). Finally, the water reaches the drip lateral, equipped with 4 L·h⁻¹ emitters (two per plant in a row of six shrubs) and an end-line plug.

The experimental deployment of the device was carried out in a dedicated irrigation sector of the Green Classroom Greenhouse (GCGH), a teaching and research greenhouse located within the Agricultural Experimentation Fields of the School of Agricultural, Food and Biosystems Engineering at the Universidad Politécnica de Madrid. The irrigation head where the system is installed is positioned on a perimeter structural wall of the greenhouse, in a space specifically designated to host hydraulic control elements. This location provides direct physical access to sensors, valves, and electronic components during the experimental phase, while maintaining their integration within a functioning irrigation environment.

The devices are installed in a semi-protected environment that combines real operating conditions with controlled experimental constraints. The distribution head is located in an area with limited space, high ambient humidity, and variable temperatures characteristic of greenhouse environments. These conditions require the use of sealed electronic enclosures, IP-rated connectors, and corrosion-resistant fittings to ensure long-term durability. Accessibility has been intentionally optimized: the assembly is elevated on the perimeter wall, allowing routine maintenance tasks (such as pressure-transducer inspection, flow meter verification, acoustic-sensor calibration, or rapid component replacement) to be performed without interrupting the irrigation schedule.

Although this installation corresponds to an experimental setup designed to validate the functionality of the device in a simple drip-irrigation system, the proposed architecture can be readily extended to more complex hydraulic networks, including larger distribution heads, multiple irrigation sectors, systems with advanced filtration, or large-scale agricultural installations. This flexibility allows the methodology and the sensors employed to be adapted to diverse operational contexts and irrigation infrastructures with greater monitoring and control requirements.

4. Conclusions

The design, development, and implementation of an embedded device, capable of being distributed along a water supply network, has been carried out, allowing the monitoring of the main relevant aspects in terms of consumption management and infrastructure status, as well as creating a tool that allows the detection of leaks in these networks.

The developed system is also able to perform a basic control over the network, allowing us not only to open and close the ducts in which it is arranged but also to establish openings and closings, partial or total, programmed with respect to temporal and physical constants obtained by the system itself.

The development of the LoRa telecommunications system allows the deployment to be viable in areas of low internet coverage which is a great advantage over other systems studied already present both in the academic and commercial field, making possible the installation in a distributed manner of several devices around a wide radius established by the Gateway.

Furthermore, the technical viability and operational reliability of the system were rigorously validated through a comprehensive battery of white-box module tests. These experimental procedures verified the correct integration of hardware and software components, ranging from signal acquisition via the ADC and digital microphone to the precise actuation of the flow control system. Notably, the field tests confirmed the robustness of the communication architecture, achieving effective LoRa transmission ranges of up to 4 km, while the successful execution of command sequences on the electromechanical valves demonstrated the device’s responsiveness. These results ensure that the prototype complies with the functional requirements necessary for deployment in operational environments.

It is therefore considered that this system is a great tool applicable in different fields. In the first place, the monitoring system capable of obtaining and combining the data of different sensors such as flow, pressure, and sound intensity, can provide enough representative data about WDN status, enabling future analysis of these variables for detecting and locating water leaks. In second place, the capability of controlling the water flow through the installed pipe is very versatile and provides the system with the capability of actuating as soon as a leak is detected, reducing the water losses and generating a great impact economically, ecologically, and socially.