Observing a Moving Target -- Reliable Transmission of Debug Logs from Embedded Mobile Devices

TL;DR

This paper evaluates various log transmission methods for resource-constrained IoT devices, highlighting the importance of application context and energy consumption, demonstrated through a case study with a self-driving slot car.

Contribution

It provides a comparative analysis of log transmission approaches tailored to IoT device constraints and presents empirical insights from a real-world case study.

Findings

Wireless transmission can meet many lifecycle phase requirements.

Energy consumption significantly impacts wireless communication reliability.

Power interruptions affect log transmission stability.

Abstract

Mobile embedded devices of the Internet of Things (IoT) face tight resource constraints and uncertain environments, including energy scarcity and unstable connectivity. This aggravates debugging, optimization, monitoring, etc.; for which logging information must be accessible throughout all phases of development and product life cycles. This work compares approaches for transmitting logs with regard to application requirements (e.g., bandwidth), resource consumption (e.g., memory), operating constraints (e.g., power supply), and the medium (e.g., UART, WiFi). A qualitative comparison suggests that the adequacy of approaches depends on the concrete application and the phase in the life cycle. We report from our case study, where the embedded mobile device is represented by a self-driving slot car (Carrera D132). With this target device, failure logs, new firmware, and monitoring data need to be exchanged. The gathered experiences support our qualitative discussion: Wireless techniques can suit the needs of many phases in the life cycle but it is particularly evident that energy consumption is crucial. With a loaded wireless stack and while transmitting logs, the car's operating voltage drop within a \~ 20 ms power interruption is \~ 1.6 times higher. This limits communication flexibility significantly.