1. Introduction
An increasing number of research studies on smart agriculture has been motivated by several challenges. First, the explosive growth of the world population [
1]. According to the United Nations (UN) Food and Agriculture Organization, an increase of up to 70% more food will be required in 2050. Second, the declining agricultural lands and exhaustion of finite natural resources such as fresh water and arable land. In addition, the decreasing number of agricultural laborers in the majority of countries. As a consequence of this agricultural workforce decline, there is an urgent need for the adoption of IoT solutions in agriculture practices to reduce the need for manual labor. IoT solutions support farmers to tighten the supply-demand gap.
Precision agriculture integrates wireless sensor networks (WSNs) with traditional agriculture to improve crop yields. This is carried out by using a large number of low-power multi-function wireless communication sensors to remotely monitor the farmland to collect environmental data, crop growth data, and livestock health data to guarantee a reduction in the possible threats to the production process and help farmers make better decisions. Recently, there is a shift from the usage of WSNs for smart agriculture to the IoT as the main enabling technology for smart agriculture. The IoT combines several technologies such as radio frequency identification, wireless sensor networks, middleware systems, end-user applications, and cloud computing.
In [
2,
3], the authors presented an IoT ecosystem architecture for smart agriculture that is made up of four main components, namely: IoT devices, communication technology, internet, and data storage and processing. First, the IoT devices are responsible for monitoring the farming environment and collecting environmental data, crop growth data, and livestock health data. Second, the role of communication technology is to establish robust, reliable, and secure communication between the cloud and the farms. Wireless communication standards are categorized based on the coverage range into short-range and long-range standards. The short-range standards include Bluetooth, near-field communication (NFC)-enabled devices, ZigBee, Z-Wave, and passive and active radio frequency identification (RFID) systems. The long-range communication standards are defined as low-power wide-area networks (LPWA). Examples of long-range communication standards include LoRa, Sigfox, and NB-IoT. The LPWA technologies provide a wide area of coverage to low-power devices [
4]. LPWA technologies outperform conventional cellular and short-range wireless technologies for different emerging smart city and machine-to-machine applications such as metering, logistics, industrial monitoring, and agriculture. However, LPWA technologies realize long-range ranging from a few to tens of kilometers and low-power operations at the expense of a low data rate. The primary aim of LPWA technologies is to achieve a 10-year battery life. LPWA technologies are suitable for delay-tolerant applications, as they achieve throughput in orders of ten kilobits per second and high latency in orders of seconds or minutes. Long range is achieved due to the use of the sub-1GHz band, and the deployment of narrow-band and spread spectrum techniques are the modulation techniques adopted by different LPWA technologies. In [
4], the authors compared the technical specifications of various LPWA technologies and standards. The choice of communication technology depended on the application of the IoT device and the type of topology. Third, the internet is the core network layer enabling the availability of data collected by IoT devices anywhere and anytime. Routing Protocol for Low Power and Lossy Networks (RPL) [
5] has been standardized by the Internet Engineering Task Force (IETF) as a routing protocol for resource-constrained nodes in IoT. RPL builds a robust topology over lossy links. A destination-oriented directed acyclic graph (DODAG) is the core of RPL, which represents a routing diagram of nodes. In [
6], the authors proposed an enhanced routing protocol based on RPL called E-RPL that decreases the number of control messages. Moreover, they proposed a flexible multi-constrained objective function (OF) that integrates several metrics such as energy, delay, and bandwidth to define the end-to-end path between the sink and a given node. The simulation results of their proposal revealed a remarkable improvement in terms of end-to-end delay, energy consumption, and routing overhead. Finally, different platforms have been developed to provide data analytics, data management, and data storage of the big data collected from sensors. Data analytics (DA) has a primary role in improving the efficiency of smart agriculture systems and in increasing productivity. DA is classified into five classes based on the requirements of IoT applications: real-time analytics, memory-level analytics, offline analytics, business intelligence-level analytics, and massive analytics. In [
7], the authors presented big IoT data analytic types, methods, and technologies for big data mining. DA can help in insurance, prediction, storage management, decision-making, farm management, and precision farming. In irrigation systems, the automated decision made by DA controls the water supply timing and quantity. The main objective of big data analytics is to analyze collected information to predict and identify recent trends, find hidden information, and finally, make decisions. Prediction, classification, clustering, and association rules are the main big data analytics methods.
Furthermore, smart irrigation is a precision agriculture application [
8] that aims to control water consumption in the agriculture sector as a scarce resource in many countries by deploying IoT technology to remotely gather information from sensors implanted in agriculture terrains to monitor soil different parameters in all stages. Based on the collected information, a decision is made on when to irrigate and the water quantity and quality required.
Table 1 reviews the recent research on IoT-based precision irrigation systems, highlighting the communication protocol, data analytics, and security techniques deployed in these systems. As can be noticed, recent research ignores security techniques and concentrates on the usage of machine learning and artificial intelligence approaches, such as fuzzy logic, artificial neural networks (ANNs), and regression models, to optimize IoT-based irrigation systems employing different environmental parameters and weather conditions to schedule the irrigation timing and the quantity of water used for irrigation. However, any alteration in the information coming from sensors and decisions passed to actuators can lead to crop damage, which is considered to be a crucial threat to the national security of any country. As such, a secure channel must be developed between the sensing layer and the decision-making entity to secure information flowing from the sensors to the decision-maker entity and to secure the decision returned to the actuators that control the irrigation system. Because precision agriculture is highly dependent on data and information from the monitored system, any alteration in such data during runtime can lead to expensive unmanageable decisions and actions from farmers. Therefore, there is a need to adopt the security mechanisms required to guarantee basic security functions: authenticity, reliability, integrity, and availability. Moreover, these security mechanisms must be lightweight to meet the requirements of constrained devices used in IoT.
In [
1], the authors reviewed the previous work conducted on smart agriculture and highlights different aspects of applying IoT solutions in smart agriculture. Moreover, the article reviews smart agriculture’s related security issues and compares security issues in the industry (urban) and agriculture (rural).
In [
16], the authors presented a classification of security threats in smart agriculture and precision agriculture environments. The authors classified security threats into six possible attacks: attacks on hardware (side channel attack and radio frequency (RF) jamming), attacks on the network equipment (denial of service (DoS), MITM (man in the middle), botnets, cloud computing attacks), attacks on data (data leakage, ransomware, cloud data leakage, false data injection, misconfiguration), attacks on applications (software update attacks, malware injection, buffer overflow, indirect attacks (SQL injection)), attacks on support chain (third-party attacks, data fabrication), and misuse attacks (cyber-terrorism, invalidation, and compliance). Another classification of security threats is underlined in [
17]. The authors classified the security requirements in smart agriculture into six challenges, namely: integrity, availability, authentication, confidentiality, privacy, and Non-repudiation, and highlighted the possible attacks under each challenge. Moreover, a review of existing solutions to IoT security problems is emphasized. In [
8], the authors added data freshness, authorization, and self-healing to the security requirements of smart agriculture. In [
18], the authors reviewed IoT communication technologies security aspects for smart agriculture. In [
19], the authors reviewed all categories of security attacks and the application of WSNs in IoT along with an evaluation of the countermeasures adopted against each type of attack.
Moreover, smart irrigation systems developed based on IoT technology consist of constrained nodes in terms of power, memory, and processing resources. As a result, conventional security protocols can not be supported in such systems [
20,
21]. Transport layer security (TLS) adds overhead in terms of memory and energy on constrained nodes. As a result of the dependence on constrained nodes in IoT-based irrigation systems, a lightweight security protocol must be deployed. Lightweight cryptography techniques balance throughput against power drain, memory usage, and gate equivalent and have lower performance when compared to cryptography standards (such as AES and SHA-256) [
22]. Characteristics of lightweight cryptography are highlighted in ISO/IEC 29192 and ISO/IEC JTC 1/SC 27. Lightweight properties are evaluated based on chip size and energy consumption and small code and/or RAM size in case of software implementation [
21]. In [
21], the authors discussed privacy in IoT in the context of developing solutions and frameworks that address profiling and tracking, localization, and tracking challenges and underlined state-of-the-art lightweight cryptographic framework for IoT. In [
23], the authors proposed a set of lightweight security protocols for encryption, authentication, and key management for IoT. The authors compared their proposed protocols with IPsec in terms of security and computational efficiency. They succeeded in achieving a decreased level of resource consumption with an increased level of security. In [
24], the authors implemented AES and PRESENT ciphers on a smartphone and provided a performance evaluation comparison between the two algorithms. AES is the symmetric block cipher defined by the National Institute of Standards and Technology (NIST) as the standard for bulk data encryption, whereas PRESENT is a symmetric ultra-lightweight block cipher that was standardized by ISO/IEC.
On the other hand, the Message Queue Telemetry Transport (MQTT) protocol has been widely deployed as an application layer messaging and information exchange protocol in machine-to-machine (M2M) communication [
25]. This is due to its ability to function with resource-constrained devices that utilize low bandwidth and unreliable links. MQTT is considered a lightweight, energy-efficient, and bandwidth-efficient communication protocol. MQTT utilizes the publish/subscribe architecture model to provide transition flexibility and simplicity of implementation. MQTT’s main components are the publishers (lightweight sensors), the subscribers (applications interested in sensor data), and the brokers (connect publishers and subscribers and classify sensor data into topics) as illustrated in
Figure 1. The data generated by a publisher are dispatched to multiple subscribers through an MQTT broker. MQTT was proposed in 1999 by Andy Stanford-Clark of IBM and Arlen Nipper of Arcom and is currently an OASIS (Organization for the Advancement of Structured Information Standards) standard; it also has a standard defined in ISO/IEC 20922: 2016.
Motivated by the increasing importance of smart irrigation systems in conserving water as a scarce natural resource, the role of precision agriculture in agriculture development, and the urgent need to apply information security techniques in the IoT part of the smart agriculture ecosystem, our aim in this research is to integrate a lightweight cryptography layer into the IoT ecosystem for smart agriculture and to investigate the deployment of a lightweight encryption protocol (the Expeditious Cipher) to create a secure channel between the sensing layer and the broker of MQTT protocol as well as between the broker and its subscribers in smart irrigation systems. This secure channel protects the sensors’ published sensitive data from eavesdropping and theft and preserves the integrity of data, in addition to protection of the decision that is made by the DA entity and returned to actuators. It should be noted that the security in IoT-based systems lies in IoT local systems consisting of devices constrained in energy and computing power.
The following points summarize the main contributions of this article.
The main contribution of this article is the integration of a lightweight cryptography layer to the IoT ecosystem for smart agriculture that meets the requirements of constrained devices used in smart agriculture in general and specifically, in our proposed IoT-based irrigation system.
The article investigates the deployment of a lightweight encryption protocol (Expeditious Cipher (X-cipher)) to create a secure channel between the sensing layer and the broker in the MQTT protocol as well as a secure channel between the broker and its subscribers in smart irrigation systems (our case study).
The proposed model is evaluated through simulation to validate the lightweight property of the chosen encryption protocol in terms of power consumption, execution time, and memory usage. Moreover, a performance comparison is carried out between the Expeditious Cipher (X-cipher), AES, and PRESENT cipher (lightweight standard protocol) in terms of power consumption, execution time, memory usage, and average throughput.
The security requirements of IoT-based agriculture systems and the potential attacks against them are discussed.
A state-of-the-art lightweight security architectures proposed for securing MQTT protocol is reviewed after highlighting the concept of lightweight cryptography.
The remainder of the article is organized as follows:
Section 2 presents our proposed secure smart irrigation system after briefly reviewing the state-of-the-art proposed lightweight security architectures for securing MQTT protocol.
Section 3 highlights and discusses the performance evaluation results of our selected lightweight encryption algorithm versus AES, in addition to a performance comparison between X-cipher and the PRESENT cipher.
Section 4 summarizes the findings of the article and highlights our suggestions for future work.