• astrocast
• cbor
• docker
• embedded systems
• hybrid networks
• internet of things
• iot mcu
• json
• machine-to-machine
• messagepack
• mobius
• non-terrestrial networks
• ntn
• onem2m
• payload optimization
• performance evaluation
• protocol buffers
• satellite iot
• scalability
• serialization
• struct zlib

The growth of massive-scale Internet of Things (IoT) systems is hindered by the verbosity of standard data formats such as JSON, particularly in hybrid networks that combine terrestrial (TN) infrastructure with constrained non-terrestrial (NT) satellite links. This thesis addresses this challenge through a comparative analysis of four device-friendly payload optimization techniques CBOR, MessagePack, a custom Struct + Zlib hybrid, and Google’s Protocol Buffers. To evaluate these methods, a realistic, containerized, and horizontally scalable client-server testbed was designed and tested under loads of up to 2000 concurrent users. The study pursued two main goals: (i) to identify which protocols could meet the strict <160-byte payload constraint of modern satellite IoT networks, and (ii) to determine the most effective overall solution in terms of throughput, latency, stability, and architectural maintainability.
Results were definitive. The initial payload analysis confirmed that only Struct + Zlib (127 bytes) and Protocol Buffers (138 bytes) achieved the necessary ~67–69% reduction to meet the NTN limit, disqualifying CBOR and MessagePack for hybrid deployments. Subsequent load testing on a four-instance server architecture demonstrated that both viable protocols sustained a 2000-user stress test with zero failures, achieving throughput above 711 RPS. A final Figure of Merit (FoM) analysis, integrating throughput, latency, and reliability, showed that Protocol Buffers (FoM: 649.5) provided a superior balance of performance and stability compared to Struct + Zlib (FoM: 633.3).
This thesis concludes that a schema-driven approach with Protocol Buffers is the optimal strategy for developing efficient, reliable, and maintainable M2M systems in hybrid TN/NT environments. Beyond technical feasibility, the findings also highlight the broader economic and cross-domain relevance of payload optimization as a cornerstone for scalable and sustainable global IoT.