Empirical Evaluation of Dynamic Task Allocation and Topological Reliability in Resource-Constrained ESP8266 Mesh Networks

Abstract

Edge computing networks rely on decentralized micro-clusters to process tasks locally, reducing latency. However, deploying these networks on highly resource-constrained hardware introduces severe bottlenecking and synchronization failures. This research investigates the practical limitations of parallel computing in edge environments by engineering a physical 3-node distributed micro-cluster using ESP8266 microcontrollers. By distributing heavy matrix multiplication tasks via a custom Wi-Fi protocol, we empirically tested Amdahl’s Law, comparing 1x versus 3x node processing speeds. The data demonstrated that network latency, rather than raw compute power, acts as the primary scaling bottleneck in extreme edge devices.

Furthermore, during sustained stress testing, a critical hardware-level anomaly was documented: a 32-bit clock underflow caused by the internal timing functions desynchronizing under heavy load, triggering catastrophic watchdog timer resets. To mitigate this, a Load-Aware Task Allocation protocol was developed, utilizing dynamic offset synchronization to stabilize the cluster and prevent chronological failure. This study proves that while data-center networking concepts can be scaled down to bare-metal microcontrollers, hardware constraints require custom firmware-level load balancing to maintain uptime. The findings provide a scalable framework for deploying fault-tolerant IoT infrastructure without relying on expensive, enterprise-grade server hardware.