Towards strengthening edge computing security through stack protection mechanisms

Stack-based attacks pose a major security risk to lightweight edge computing devices used in smart agriculture. This study investigated a secure stack-protection model for Advanced RISC Machines (ARM) Cortex-M4 microcontrollers commonly used in agricultural Internet of Things (IoT) workloads in Nigeria. Six configurations were simulated: Memory Protection Unit (MPU), Stack Canary, Shadow Stack under Mask (SuM), Control-Flow Integrity (CFI), a Hybrid MPU--SuM mechanism, and an unprotected control. The Python-based simulation modelled 500 sensor readings and 100 simulated attack attempts for each mechanism. The Hybrid mechanism prevented all simulated attacks within the defined threat model while incurring only 2.90% performance overhead and a 2.83% reduction in estimated battery life. It maintained an operational lifetime above 5.8 months on a 2000 mAh battery. Although CFI also achieved complete prevention in the simulation, its 10.00% overhead reduced its suitability for latency-sensitive and duty-cycled sensor workloads. The results indicate that hardware-assisted protection, particularly MPU enforcement and the Hybrid MPU--SuM approach, can provide practical stack-level security for resource-constrained agricultural IoT nodes without requiring additional hardware.