Throughput bottleneck shift in an embedded Wi-Fi gateway

Abstract

An ESP32 software access point (SoftAP) often serves as the sole transmission control protocol (TCP) gateway in field deployments without routers; however, prior benchmarks varied only one factor at a time. We varied two factors that differ in practice: TCP connection count (3, 5, and 10) and per-packet payload size (10, 64, and 256 bytes), yielding 9 configurations. Each was tested at in creasing send frequencies, with each frequency repeated 3 times, for a total of 246 60-second runs. Two-way analysis of variance (ANOVA) on the 30 Hz data confirms that both factors and their interaction are significant (all p < 0.001). With 10 connections and 10-byte payloads, the gateway sustained 3,237 pack ets per second (pps) with zero loss; switching to 256-byte payloads dropped the safe rate below 946 pps and pushed central processing unit (CPU) utilization past 100%. In our experiments, system constraints shift from per-frame media access control (MAC) overhead to channel-bandwidth saturation as payload size increases. We present a stability-region chart that helps designers understand the limits of system stability, optimize system performance, and choose the appro priate connection count and message size.