Agricultural production represents a strategic sector of any national economy. This is particularly true in Latin American countries, where traditionally a large percentage of the population is rural and depends on traditional agricultural production to live. Efficiently monitoring crops is critical because it significantly increases production, rationalizes the use of water and other consumables, and produces value-added crops [1]. However, crop monitoring can sometimes present technological difficulties, thus increasing operational costs and maintenance. The cost and maintenance of complex systems often exceeds what smaller farmers can invest. Smaller producers in developing countries cannot exploit the benefits of a scale economy. Unfortunately, because smaller, traditional producers often cannot compete, they abandon their fields seeking improved economic opportunities in the cities. In Latin America countries, approximately 40% of the population has migrated from rural to urban settings either in their own countries or abroad [2]. This migratory trend has resulted in decreased agricultural production and increased agricultural imports, leading to increased trade deficits and foreign dependency on foreign sources, both of which can contribute to making basic products difficult to acquire [3].

Mexico is not an exception to Latin American trends regarding its agricultural sector. The modernization of farming practices and the use of technology in Mexico’s fields has been at the center of great debate. Because most of Mexico’s land is not suitable for traditional agriculture, modernization of agricultural practices and the principle of competitive advantage infer a transition, in part, to non-traditional crops cultivated with emerging technologies, including wireless technologies, such as sensors and actuators. This automation may not only significantly improve production and crop quality, but more efficiently use often scarce natural resources such as soil and water.

Precision agriculture techniques, whose objective is to efficiently use consumables such as fertilizers, pesticides, soil, and water, among others, can be applied in both open and closed spaces. An additional advantage of precision agriculture techniques is that they can reduce the use of dangerous agricultural products that contaminate the environment. Precision agriculture traditionally involves global positioning (GPS) to help identify problems related to ground monitoring, insect pests, humidity and crop density, among others. Present satellite technology and image analysis, however, can be rather costly and imprecise, as many problems are too small to be detected by satellite imagery.

Sustainable agricultural practices emphasize the development of biotechnology, techniques to increase crop production, and the application of technology to agricultural production, among others. However, these practices make applying state-of-the-art technology in many parts of the world difficult to achieve. Expanding the role of technology to monitor and control crops and otherwise automate agricultural practices is essential to decrease costs and provide benefits to a greater percentage of producers worldwide. Precision agriculture, comprised of sensors, wireless networks, computer hardware, software applications, and wireless communication technologies can significantly reduce the time a producer requires to make important decisions related to resource management, planning, administration, process analysis and evaluation, which ultimately contribute to improved decision making.

Presently, the agricultural sector does not sufficiently employ technology and informatics to modify production practices. Although progress is being made with regards to the deployment of sensors, wireless networks, actuators and other electromechanical devices in agricultural settings, there are still important areas of development that have not been sufficiently explored. The digital divide also affects agricultural practices in developing countries, as many current innovations have not yet “filtered down”. Embedded systems and wireless technologies can, in the long run, reduce costs and increase profits in countries with favorable year-round climates that permit multiple harvests but lack other essentials required to maximize their potential.

Advances in micro-electro-mechanical systems (MEMS) technology have made the deployment of wireless sensor nodes a reality, in part, because they are small, inexpensive and energy efficient. Each node of a sensor network consists of three basic subsystems: a sensor subsystem to monitor local environmental parameters, a processing subsystem to provide computational support to the node, and a communication subsystem to provide wireless communications to exchange information with neighboring nodes. Because individual sensor nodes can only cover a relatively limited area, they need to be connected to one another in a coordinated manner to form a wireless sensor network (WSN), which can provide large amounts of detailed information about a given geographic area. Consequently, a wireless sensor network can be described as a collection of intercommunicated wireless sensor nodes which coordinate to perform a specific action. Unlike traditional wireless networks, WSNs depend on dense deployment and coordination to carry out their task. Wireless sensor nodes measure conditions in the environment surrounding them and then transform these measurements into signals that can be processed to reveal specific information about phenomena located within the coverage area surrounding these sensor nodes.

However, the imperative necessity to control physical variables such as temperature, relative humidity, soil moisture, etc., has led to the development of wireless sensor and actor networks (WSANs), which are commonly composed of heterogeneous devices referred to as sensors and actuators. Sensors are low-cost low-power multi-functional devices that communicate wirelessly for short distances. Actuators are usually resource-rich devices with greater processing capabilities, higher transmission capabilities, and longer battery life. Actuators collect and process sensor data and perform specific actions within a specified environment based on the information they receive.

Future applications will extensively employ wireless sensor networks that function in real time in conjunction with communications systems, mechanical actuators, and even robots to monitor and intervene in crop cultivation. A wireless sensors network (WSN) permits remote monitoring of many parameters, depending on the type of sensors used and the coverage area. This type of network consists of a large number of sensor nodes that are wirelessly connected to each other, to electromechanical devices, and to a communications network, all of which form a triad to monitor and control crop development. Generally, each node of a WSN consists of sensors and/or actuators. Sensors are characterized by their limited memory and computation capacities, but one advantage of sensors is that they require little power to perform their functions. Wireless sensor networks consisting of many nodes are currently being used in densely populated large scale areas. WSNs can have homogenous structures, where all nodes present similar characteristics, or heterogeneous structures, where some nodes are more powerful than others or are differentiated by physical characteristics, including the type of battery or antenna the individual nodes use, or whether specific nodes are static or dynamic.

WSNs have a variety of applications. Examples include environmental monitoring—which involves monitoring air, soil and water, condition-based maintenance, habitat monitoring, seismic detection, military surveillance, inventory tracking, smart spaces, etc. [4,5]. Despite their many diverse applications, WSNs pose a number of unique technical challenges because of fault tolerance (robustness), scalability, production costs, operating environment, sensor network topology, hardware constraints, transmission media and power consumption.

In this work, we selected a 100-meter row of watermelons in a 6-hectare field, where we placed sensors linearly at 5-meter intervals. Watermelons were chosen as the crop for study because they require a specific temperature and humidity to optimally ripen. However, our proposed technological platform can also be applied in other precision agriculture applications because it incorporates a modified version of LORA-CBF, a wireless location-based routing algorithm that uses cluster-based flooding, which, as mentioned previously, addresses the scalability concerns of wireless sensor networks, including monitoring of residual battery energy.

The remainder of this paper is organized as follows: Section 2 classifies routing algorithms and their application in wireless sensor networks. Section 3 describes the technological platform for wireless sensor nodes. Section 4 reviews the proposed hierarchical and flat algorithms for our wireless sensor network. Section 5 explains the system evaluation. Section 6 describes the evaluated scenario and the results obtained from the small-scale network and Section 7 summarizes our work and proposes future research.

Routing protocols for wireless sensor networks can be classified as data-centric, hierarchical or location-based.

The design of the agricultural platform system faced three significant challenges: building a sufficiently lightweight, energy efficient hardware capable of monitoring and control physical variables; incorporating and evaluating different operating systems and algorithms into a software to achieve autonomous transfer sensing and control variables; and integrating subsystems such as microprocessor, sensors and actuator modules and wireless networking into a fully functional wireless platform solution.

Location Routing Algorithm with Cluster-Based Flooding (LORA_CBF) [32] was developed by the principal author of this work, and was modified to meet the requirements of precision agriculture applications.

LORA-CBF is formed with one cluster-head, zero or more members in every cluster and one or more gateways to communicate with other cluster-heads. Each cluster-head maintains a “Cluster Table.” A “Cluster Table” is defined as a table that contains the addresses and status of neighbor nodes. A node in LORA-CBF can be in any of the following four states:

Undecided: A node is in this transitional state when it is in search of a cluster-head. Nodes are initially undecided when they enter the network or when they wake up.

The following metrics were used to evaluate the LORA-CBF algorithm in a hierarchical and a flat architecture.

Route discovery time (latency): the amount of time the source has to wait before sending the first data packet.

In the Figure 5, the red line represents the hierarchical architecture and the blue line the flat architecture. Figure 5(a) shows that the latency of the flat algorithm increases linearly in proportion to the number of hops. On the other hand, the latency of the hierarchical algorithm increases irregularly because of its group formation mechanism. However, it is important to note that regardless of whether the algorithm is flat or hierarchical, latency increases proportionally to the number of hops. Figure 5(b) shows the Packet delivery ratio. In this scenario, both the flat and hierarchical algorithms behave the same. Figure 5(c) shows that End-to-End delay increases proportionally for both algorithms, although the flat algorithm’s End-to-End delay increases linearly while the hierarchical algorithm’s increase in End-to-End delay is more irregular due to its group formation mechanism. This mechanism substantially increases End-to-End delay. Figure 5(d) represents the throughput which is significantly better in the hierarchical algorithm because of its superior packet transmission mechanism, which substantially reduces the possibility of collisions. Figure 5(e) shows the routing load. The flat algorithm performs better than the hierarchical algorithm with regards routing load because it does not have a group formation mechanism that can increase packet transmission. Figure 5(f) shows overhead (packets) which is the total number of routing packets that are generated divided by the total number of data packets transmitted, plus the number total routing packets. Again, the flat routing algorithm performs better than the hierarchical algorithm with regards to overhead.

In summary, the hierarchical architecture adds more delay in terms of route discovery time, End-to-End delay, routing load and overhead, but it significantly improves throughput. The main advantage of using the hierarchical architecture vs. the flat architecture is network scalability. The hierarchical architecture is more scalable and can handle a greater number of nodes.

The scenario in Figure 6 was used to evaluate the proposed routing algorithms. Five wireless sensor nodes were employed to represent a small-scale network. Node N2 represents the node which starts the routing process.

The wireless sensor network consisted of 20 sensors that were placed linearly at 5 meter intervals. The position of the four wireless sensor nodes is shown in Figure 6. The temperature and humidity sensors were placed on wooden rods placed in the soil. We chose to place the nodes in this way because we wanted the average air temperature and humidity at the intermediate height between the plants. The soil moisture and temperature sensors were placed approximately 5 cm from each other at each 5-meter interval. The sensors were then placed just in the soil a few millimeters below the actual ground level.

Each wireless sensor node has four ports, so we connected three soil moisture and temperature sensors and one humidity and temperature sensor per wireless sensor node sensing each one of the variables every hour. The wireless sensor network was deployed on Monday 13/12/2010 at 13:00, approximately, and was recovered on Wednesday 15/12/2010 at 12:00.

In this work, a new platform for wireless sensor networks including its embedded operating system and its routing algorithm was evaluated in terms of route discovery time, packet delivery ratio, End-to-End delay, Throughput, routing load and overhead. A flat and hierarchical algorithm was evaluated in a small-scale network under test bed conditions in a watermelon field. The flat algorithm proved to be superior with regards to route discovery time, End-to-End delay, and routing load and overhead. The hierarchical algorithm proved to be superior regarding Throughput and scalability. In small-scale network applications, we found that the flat algorithm is more suitable because of its simplicity. Results show that LORA_CBF is suitable for both flat and hierarchical algorithms and is suitable for small-scale agricultural use. We conclude that our proposed technological platform with a modified version of the LORA_CBF routing algorithm can also be applied to precision agriculture because it is a wireless location-based routing algorithm that uses cluster-based flooding and monitors residual battery energy.