The design and development of a wireless sensor network for soil moisture measurement in an unlevelled 10 km × 10 km area, is described. It was specifically deployed for the characterization of a reference area, in campaigns of calibration and validation of the space mission SMOS (Soil Moisture and Ocean Salinity), but the system is easily extensible to monitor other climatic or environmental variables, as well as to other regions of ecological interest. The network consists of a number of automatic measurement stations, strategically placed following soil homogeneity and land uses criteria. Every station includes acquisition, conditioning and communication systems. The electronics are battery operated with the help of solar cells, in order to have a total autonomous system. The collected data is then transmitted through long radio links, with ling ranges above 8 km. A standard PC linked to internet is finally used in order to control the whole network, to store the data, and to allow the remote access to the real-time data.
Ground humidity and its space-temporal evolution are very important in climatic and prediction models and they must be taken into account in the monitoring of the hydrology and the vegetation. In this sense, soil moisture maps are very powerful tools that can be used in a huge range of applications, like desertification studies and, indirectly, global climate change studies [
The Soil Moisture and Ocean Salinity (SMOS) space mission from the European Space Agency (ESA) [2,3] launched a mini satellite November the 2nd, 2009, with first data received November 20th. This satellite is equipped with an L-band microwave interferometric radiometer. The aim is to observe soil moisture over the continents and sea surface salinity over the oceans with resolution enough to be used in global climate studies. So collected data need to be validated with field measurements, made with specific sensors providing direct soil moisture measurements. Therefore, extensive validation and calibration areas are needed.
This paper describes the design, development and implementation of an automated wireless network of soil moisture sensors using a network technology over a homogeneous area of 10 km × 10 km that is used in SMOS calibration/validation activities.
The network is developed on one of the primary validation areas for SMOS land data and products. More specifically, this area is located in the Utiel-Requena plateau, 80 km to the west from Valencia (Spain), falling into the area of influence of the “Valencia Anchor Station” [
The predominant crops of the zone are grapevines, almonds trees, olive trees and shrubs. The soils are mainly Calcisol (CL) and Cambisol (CM), with the subclasses Haplic Calcisol (CLH), Petric Calcisol (CLP) and Calcaric Calcisol (CLC). A detailed thematic map (
is statistically different between units and homogeneous inside each one. This way, a representative soil moisture value of each area unit can be obtained sensing this parameter in just one point inside the unit. This methodology allows using a low number of sensors for covering the 10 km × 10 km area. Specifically, we will consider zones with vineyards, fruit trees and shrubs for locating the soil moisture sensors in order to get representative data. Consequently, position as well as number of sensors will be conditioned by both the soil type and use. As a trade off, we decided to install 12 sensing stations, with the following distribution: 5 sensors into vineyard lands (V1, V2, V3, V4 and V5), 3 sensors in fruit tree fields (T1, T2, T3) and 4 sensors over shrub zones (M1, M2, M3, M4). This way, all possible combinations of soil type/soil use are covered. Obviously, new nodes could be easily added to the network, if necessary.
The selection of the transmission mechanism is hardly conditioned by the characteristics of the environment where a sensor network is developed [
The communication hardware consists, basically, of:
In order to modulate/demodulate the digital signal and then transmit it via radio, a transmitting/receiving commercial (Radiomodem Wlink8S from DMD [
For increasing the range of the links, we are forced to use external antennas. Two types of antennas are chosen according to their radiation pattern. Both will be used depending on the position of the particular node. We used omnidirectional antennas with a gain of 7.3 dBi (GP-901, by SIRIO) and directional Yagi antennas with a gain of 12.0 dBi (SY-910, by SIRIO) [
Once the radio frequency module and the antennas were chosen, we estimated the respective link ranges, paired considered, assuming conditions of free space and direct vision between antennas, and without considering several attenuations like obstacles, meteorological inclemencies, interferences, diffractions, etc. [
The final network implementation needs to take into account the defined units, the zone relief and the maximum ranges previously between links. Taking advantage of a small mountainous elevation with direct vision of the most of the region, and after considering different possible topologies [
Each link was individually calculated and measured. The summary of the main parameters of the link tests are shown in
After choosing the communication system, deciding network topology and fixing the position of the nodes, the hardware and software of each one and the development of the communication protocol is now explained. Each node is equipped with necessary electronics in order to provide soil moisture measurements, store the acquired data and control the communication with other sensors.
A simplified block diagram of a sensor node is shown in
The nodes need to be independently fed by means of solar cell operated batteries, due to the difficulty of accessing the mains in every node of the network. Specifically, we have made use of a MSX 01F (BECOsolar) solar cell (1 W of nominal power) [
The soil moisture was measured by means of a ThetaProbe ML2x (Delta-T Devices) [
humidity of the soil. The conversion relationship depends on the soil characteristics and needs to be calibrated for any particular case. The probe can be fed with a voltage between +5 V and +15 V. Because the probe does not need to be continuously measured, it is connected/disconnected through a typical common emitter transistor.
For monitoring medium-scale sensor networks, a microcontroller based architecture has been proved to be powerful enough [
(1) Acquisition of the measured data, provided by the soil moisture sensor, through an A/D port. The differential voltage needs to be previously converted to a 0 V/+5 V level by means of a specific instrumentation amplifier together with an operational amplifier;
(2) Acquisition of the ambient temperature, through a digital integrated thermometer (DS1631) connected to the I2C bus;
(3) Monitoring of the battery charge voltage status through an additional A/D port;
(4) Communication with the RF module, making use of the Universal Asynchronous Receiver Transmitter (UART). For this purpose a RS-232/TTL level converser (MAX3222, Maxim) equipped with a shutdown (low consumption) mode was specifically designed and implemented;
(5) Synchronization of the network, with an external Real Time Clock (RTC), controlled through the I2C bus. We have selected the DS1678 (Maxim). The RTC also interrupts periodically the microcontroller, with a configurable alarm, to start a new measurement.
For the data storage, a 512 kbit I2C controlled external EEPROM is utilized. If a particular data ratio of a measurement every ten minutes is considered, this memory allows a backup communication failure protection of 7 days in the worst case (if the failure is of the central node) or a maximum of 84 days (end node failure).
For a proper function of the network, all nodes must be synchronized in order to correctly correlate the measurements as well as to optimize the communication protocol by reducing unnecessary waiting times. In the network, each node will be identified by a unique software address. These nodes should operate in a standby (low power consumption) mode when neither measurement nor transmission activity is required. The whole network will be regulated by a PC, directly connected to the central node. This particular node will initialize the network, synchronize it, and store the data collected from all the nodes. Additionally, an error detection/correction protocol is also considered in this communication system.
In this sense, a Medium Access Control (MAC) based protocol [
Commands fall in one of the following categories:
(1) Transmission acknowledgement (ACK/NACK);
(2) Shutdown activation/deactivation;
(3) Measurement start;
(4) Data asking;
(5) Data sending;
(6) Synchronization;
(7) Reset.
Initially, the network is in a standby state. A personal computer, directly connected to the central node, initializes the network. Then, synchronization and start measurement commands are sequentially sent to the network. These commands travel hierarchically through the network until reaching all the nodes. Sometime after, and in the same way, the computer sends commands asking for data. Then, the stored data from the performed measurements travel back from the end nodes to the central one. All the data are then organized by the PC, allowing the subsequent analysis.
At a certain time a given node receives a synchronization order and a start measurement command. In response, the microcontroller configures the RTC and states an alarm that will be activated each time a measurement must be made. This way the node is periodically capturing the defined measurements and storing them in the EEPROM, together with the clock information. At any time, a node can receive a command asking for data. In this case, the node will send the stored data to the upper level node. For clarity, In
The protocol is energy optimized by properly using the shutdown mode. In this sense, the communication time is highly minimized, leading to considerable energy savings. The averaged power consumption can be estimated as in [
measurements over the considered region, and taking into account the characteristics of the utilized solar cell, the averaged charge of the battery of a node can be estimated to be about 480 mAh. From data in
The final node implementation is depicted in
The box is then attached to a 3-meter mast, also holding the antenna, and close to it. The particular arrangement and final installation configuration hardly depends on the ground conditions, as we try to show in
The diagram of the overall system is shown in
As easily observed, the soil moisture (
We have successfully designed and implemented a completed sensor network for soil moisture monitorization using radio links in a wide 10 km × 10 km reference area. By sampling each defined unit, a low density of sensors has been achieved. This fact has allowed us to cover a higher area with lower cost of implementation.
The network, once configured, is automatic. It makes use of radiofrequency 868 MHz RF modules with a hierarchical configuration capable to cover the whole area with radio-links up to about 7 km.
Each node is equipped with a soil moisture probe and a microcontroller based electronics board in order to capture and store the environmental data, at the desired intervals. A solar cell supplies the power to a rechargeable battery, allowing the system to be autonomous.
A central node, directly connected to a PC, is dedicated to configure and control the network, as well as to collect the data from all the nodes and to allow their access through Internet from anywhere. This way, we have completed a typical sensor network including a real time web application to access the soil moisture data.
The system is fault tolerant, and immediately extensible to a higher number of nodes in a wider region, or to the measurement of additional environmental variables.