Grant: Discovery Grants Program - Individual

The global population is expected to reach 10 billion by 2050, which will increase the food demand by 70% [3]. Unfortunately, the current changing climate and poor agricultural practices have prevented food production from growing at the same rate, and more than 800 million people are already facing food insecurity [1]. Data-driven (or smart) farming is widely recognized as a disruptive force towards improving global food systems. The global smart farming market value is expected to be 33 billion by 2027 due to the adoption of digital technologies like the Internet of Things (IoT), big data analytics, AI/ML, and edge/cloud computing [48]. These technologies allow farmers to address issues like low crop yields, pest diagnostics, and climate adaptation [42-44]. Despite the increasing availability of digital technologies in farms worldwide, current smart farming systems are still lack trusted designs capable of solving issues beyond their basic functional requirements. For example, existing systems [4, 20] neglect any security, privacy, robustness, energy-efficiency and/or usability concerns farmers may have. Thus, brittle and ad hoc designs tend to hinder systems' adoption which delays the pressingly needed digitization process in agriculture. Solving this issue requires re-thinking smart farming systems from an end-to-end perspective, i.e., from sensor operation at the edge to deep analytics in the cloud, considering agriculture-specific constraints, which is missing in existing works. It is necessary to consider that rural areas typically face long power outages, have poor connectivity compared to urban areas, and take time to produce meaningful datasets due to seasonal crop lifecycles. The long term vision of this proposal is developing secure, robust, and energy-efficient smart farming systems with feasibility and efficiency demonstration. To that end, we will incorporate recent network programmability and artificial intelligence advances into the systems' design while addressing important research questions (short term objectives): i) can attackers leak sensitive farm information from crop data? How can farmers control the security/privacy of their systems? ii) how can we make smart farming systems more robust by using synthetic data? and iii) how to reduce the energy consumption of AI/ML models for smart farming? The proposed research will benefit Canadian technology, food, the environment, and the economy. It will train HQPs required for realizing secure and sustainable smart farm systems, including large AI/ML models essential in data-driven decision-making for efficient and sustainable food production. This objective aligns with Canada's sustainability movement. Specifically, HQPs will learn necessary skills in cybersecurity, AI/ML model development, data analytics, and Big Data systems, highly demanding areas in the Canadian job market in industry and academia [46-47].