Performance Evaluation of MQTT, CoAP and HTTP Protocols for Application-Specific IoT Scenarios

Abstract

The proliferation of the Internet of Things (IoT) has introduced significant challenges in protocol selection for resource-constrained devices operating in smart healthcare, industrial automation, and agricultural monitoring. While lightweight protocols like Message Queuing Telemetry Transport (MQTT) and Constrained Application Protocol (CoAP) are designed for low-power Wide Area Networks, the ubiquity of Hypertext Transfer Protocol (HTTP) persists despite its substantial overhead. This research provides a multi-dimensional performance evaluation of MQTT, CoAP, and HTTP using the NS-3 discrete-event simulator. The study uniquely benchmarks these protocols across nine performance metrics, including Average Latency, Network Jitter, CPU Energy Usage, and Network Footprint, under varying iteration loads (50 and 500) to simulate both nominal and high-stress IoT environments. Our empirical analysis demonstrates that while CoAP provides the highest throughput and lowest resource consumption, MQTT offers superior reliability in persistent monitoring scenarios. Notably, the evaluation reveals a critical scalability threshold for HTTP, where high-stress iterations lead to a complete communication collapse (0% PDR) due to TCP-induced network saturation. By correlating system-level impact (CPU usage) with network-level stability (Jitter), this paper proposes an application-specific decision framework. The findings serve as a technical roadmap for optimizing next-generation IoT deployments based on specific latency and energy requirements.

Introduction

The rapid expansion of the Internet of Things (IoT) has created a need for efficient communication protocols for billions of resource-constrained devices operating in lossy, bandwidth-limited networks. The primary protocols considered are:

• MQTT (Message Queuing Telemetry Transport): Broker-based, publish-subscribe, TCP connection; state-aware and persistent.

• CoAP (Constrained Application Protocol): Lightweight, request-response, UDP-based; optimized for Machine-to-Machine (M2M) communication.

• HTTP (Hypertext Transfer Protocol): Traditional client-server over TCP; universally compatible but high overhead for constrained devices.

Research Gap

• Previous studies focused on limited metrics under ideal network conditions.

• Few studies examined energy efficiency, high-stress scenarios, or multi-dimensional correlations between network-level (e.g., jitter) and system-level (CPU energy) metrics.

• There is a lack of comprehensive benchmarking under high-load iterations, especially for stress testing large-scale IoT deployments.

Methodology

1. Simulation Environment:

   • NS-3 discrete-event simulator for realistic wireless IoT networks.

   • IEEE 802.11b in Ad-hoc mode with constant node positions.

   • Cluster-based topology with nodes communicating to a centralized gateway.

2. Experimental Scenarios:

   • Low-load: 50 iterations (baseline network stress).

   • High-load: 500 iterations (stress test for network saturation).

3. Performance Metrics (9 Total):

   • Average Latency: Time for packets to travel from sender to receiver.

   • System Throughput: Data successfully transmitted per second.

   • Bandwidth Overhead: Ratio of control data to payload.

   • Packet Delivery Ratio (PDR): Reliability of delivery.

   • CPU Energy Usage: Proxy for computational energy consumption.

   • Network Jitter: Variation in end-to-end delay.

   • Packet Loss Rate: Frequency of dropped packets.

   • Network Footprint: Total data transmitted for task completion.

   • Scalability: Performance change between 50 and 500 iterations.

Key Findings

• CoAP: Performs best in high-speed, low-resource, and high-stress environments due to its UDP-based lightweight architecture.

• MQTT: Offers stable performance under intermittent connectivity, but higher CPU usage and latency under extreme stress.

• HTTP: Suffers catastrophic failure under high-load conditions (100% packet loss, extreme latency), making it unsuitable for constrained or high-stress IoT deployments.

Contributions

1. Multi-Dimensional Benchmarking: Evaluates protocols across nine metrics for a holistic view of network and system performance.

2. Stress Testing & Scalability Analysis: Compares low and high-load scenarios to simulate real-world IoT deployments.

Conclusion

This research presented a comprehensive performance evaluation of three widely used IoT communication protocols—MQTT, CoAP, and HTTP—based on nine critical performance metrics under constrained network conditions. The experiments were conducted under varying communication loads ranging from 50 to 500 iterations to simulate both normal and high-stress IoT environments. The results indicate that CoAP demonstrates superior overall performance in constrained environments by achieving the lowest latency (0.55 ms), the highest throughput, minimal bandwidth overhead, and the most efficient CPU utilization (8.5%). These characteristics make CoAP highly suitable for energy-sensitive and delay-critical IoT applications. MQTT exhibited excellent reliability, consistently maintaining a 100% Packet Delivery Ratio (PDR) across all scenarios. Its broker-based publish–subscribe architecture makes it particularly suitable for applications that require persistent, state-aware communication and continuous data monitoring. In contrast, HTTP proved to be unsuitable for high-load IoT scenarios in constrained environments. The protocol experienced complete communication failure during high-frequency data transmission, resulting in 100% packet loss due to its significant header overhead and connection establishment mechanisms. Overall, the study concludes that while MQTT is preferred for reliable and continuous event-driven communication, CoAP emerges as the optimal protocol for resource-constrained, battery-operated IoT nodes that demand high-speed and energy-efficient data transfer.

References

[1] M. Elhoseny, A. K. Sangaiah, and K. Shankar, \"Internet of Things (IoT) for Next-Generation Smart Systems: A Review of Current Challenges, Future Trends and Prospects for Emerging 5G-IoT Scenarios,\" IEEE Access, vol. 8, pp. 117600–117620, 2020. [2] S. Naik, V. Maral, and M. A. Al-Qutayri, \"MQTT Protocol for the IoT,\" in Proc. International Conference on Advanced Communication Technologies and Networking (CommNet), 2019, pp. 1–6. [3] R. Khan, S. U. Rehman, and I. Khan, \"A Reliable CoAP Protocol for IoT Communication,\" in Proc. International Conference on Information Networking (ICOIN), 2021, pp. 318–323. [4] A. Aijaz and A. H. Aghvami, \"Performance of IoT Protocols over Constrained Wireless Networks,\" in Proc. International Conference on Computer Networks (CN), 2021, pp. 45–52. [5] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari and M. Ayyash, \"Internet of Things: A Survey on Enabling Technologies, Protocols, and Applications,\" IEEE Communications Surveys & Tutorials, vol. 17, no. 4, pp. 2347-2376, 2015. [6] Z. Shelby, K. Hartke and C. Bormann, \"The Constrained Application Protocol (CoAP),\" RFC 7252, 2014. [7] A. Banks and R. Gupta, \"MQTT Version 3.1.1,\" OASIS Standard, 2014. [8] R. Fielding et al., \"Hypertext Transfer Protocol -- HTTP/1.1,\" RFC 2616, 1999. [9] J. Dizdarevi?, F. Guzman, B. Salazar and R. Mariano, \"A Survey of Communication Protocols for IoT: HTTP, CoAP, MQTT and AMQP,\" Ad Hoc Networks, vol. 88, pp. 173-195, 2019. [10] D. Thangavel et al., \"Performance evaluation of MQTT and CoAP via a common middleware,\" 2014 IEEE ISSNIP, pp. 1-6, 2014. [11] B. Mishra and A. Kazi, \"A Comparative Study of MQTT and CoAP Protocols in IoT,\" International Journal of Computer Applications, 2023. [12] G. Gagliardi et al., \"Performance Evaluation of IoT Protocols in Real-Time Healthcare Monitoring,\" Sensors, vol. 23, no. 4, 2023. [13] N. Naik, \"Choice of effective messaging protocols for IoT systems: MQTT, CoAP, AMQP and HTTP,\" 2017 IEEE Systems Engineering Symposium (ISSE), pp. 1-7, 2017. [14] S. Singh and M. Kumar, \"Stress Testing IoT Protocols: A Comparative Analysis under High Traffic,\" Journal of Cloud Computing, 2024. [15] R. Patel and S. Jha, \"Impact of Packet Loss on MQTT and CoAP Reliability: An NS-3 Study,\" IEEE Access, 2024. [16] K. Grgi? et al., \"Performance Evaluation of HTTP, CoAP and MQTT Protocols,\" 2016 IEEE sInternational Conference on Smart Systems, 2016. [17] M. Collina, G. E. Corazza and A. Vanelli-Coralli, \"Introducing the QEST broker: HTTP / MQTT bridge for the Internet of Things,\" 2012 IEEE 23rd PIMRC, 2012. [18] H. Wang and Z. Li, \"Cross-Layer Performance Optimization for IoT: MQTT vs CoAP,\" IEEE Transactions on Industrial Informatics, vol. 21, 2025. [19] B. Pradeep Kumar et al., \"Comparative Study of IoT Protocols: MQTT, CoAP, and HTTP,\" International Journal of Advanced Trends in Computer Science, 2019. [20] Y. Chen, \"Latency and Jitter Analysis in UDP-based IoT Protocols,\" IEEE Access, vol. 11, 2023. [21] R. Marti, \"Evaluating Energy Consumption of MQTT and CoAP in IoT,\" IEEE Internet of Things Journal, 2016. [22] M. Hassan, \"Evaluation of HTTP/3 and its comparison with MQTT for future IoT,\" Scientific Reports, 2025. [23] H. Al-Kashoash, \"Congestion Control Mechanisms in CoRE and MQTT Networks,\" Elsevier Ad Hoc Networks, 2024. [24] J. Torres, \"Analytical Model for Energy Consumption in Battery-Powered IoT Nodes,\" IEEE Sensors Letters, 2024. [25] S. Khan et al., \"Benchmark Analysis of CoAP vs MQTT in Industrial IoT Scenarios,\" MDPI Sensors, vol. 25, no. 2, 2025. [26] B. Mishra, \"CPU Usage and Memory Footprint of Application Layer Protocols,\" Int. J. of Comm. Systems, 2023. [27] L. Z. Yan, \"Scalability Analysis of IoT Protocols in Large-Scale NS-3 Simulations,\" Journal of Network and Systems Management, 2024. [28] M. Abdel-Basset, \"Scalability Challenges in MQTT-driven Smart Cities,\" Sustainable Cities and Society, vol. 98, 2024. [29] A. Larmo et al., \"Evaluation of MQTT and CoAP for Remote Monitoring,\" IEEE Communications Magazine, 2019.

Copyright

Copyright © 2026 Mahak Mishra, Shiva Guru. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.