MESHMIND: DECENTRALIZED LAN-BASED COORDINATION FOR LARGE LANGUAGE MODELS: DESIGN, IMPLEMENTATION, AND EVALUATION
DOI:
https://doi.org/10.4314/njt.2026.6068Keywords:
Large Language Models (LLMs), Edge Intelligence, Distributed Computing Framework, Collaborative Model Processing, Decentralized Artificial Intelligence Systems, Peer-to-Peer NetworkingAbstract
Big developments in large language Models (LLMs) have increased usage of smart (intelligent) apps but have also made it more dependent on cloud infrastructures that have limitations in privacy, latency, costs and network dependence. This paper will propose a new P2P Framework called "MeshMind". MeshMind is a fully decentralized peer-to-peer (P2P) URL Application that provides secure, collaborative access to the LLMs for the purpose of providing extensive access to computational resources or services within LANs. MeshMind allows users to collaboratively utilize multiple heterogeneous devices and enables a distributed architecture for LLM processing. This paper details how a hybrid communication structure incorporates UDP for automatic peer discovery and TCP for establishing reliable synchronized communications, transferring files and updating capability status. Finally, MeshMind employs a multilayered security mechanism, including the use of HMAC- signed protocol messages and JWT-based session control. Ollama-enabled devices can act as hosts for LLMs, while lighter weight devices, referred to as peers, and may utilize authenticated delegation to offload inference requests according to workload heuristics that adapt based on real-time conditions. Experimental evaluation conducted using a multi-device local area network (LAN) test bed showed that MeshMind successfully discovered peers, synchronized devices reliably, and distributed inference processing in an ultimately scalable architecture, resulting in sub-3 second latencies as the network(s) increased in size. The throughput improved significantly with the increasing number of peers while resource usage was evenly distributed across all peer devices by intelligently distributing workload, resulting in significantly improved processing capabilities without increasing costs associated with using additional hardware resources. Mathematical simulations of MeshMind performance confirmed the overall efficiency, high level of network stability, and ability to achieve a near-linear scale within the distributed architecture.
References
[1] Maryam, V. and Fotouhi, H. "Federated learning at the edge in Industrial Internet of Things: A review." Sustainable Computing: Informatics and Systems, 46(1), pp.101087, 2025. https://doi.org/10.1016/j.suscom.2025.101087
[2] Wei, W. and Liu. L. "Trustworthy distributed AI systems: Robustness, privacy, and governance", ACM Computing Surveys 57(6), pp. 1-42, 2025. https://doi.org/10.1145/3645102
[3] Zhang, M., Shen, X., Cao, J., Cui, Z., and Jiang, S., "Edgeshard: Efficient LLM inference via collaborative edge computing." IEEE Internet of Things Journal, 12(10), pp.13119-13131, 2024. doi: 10.1109/JIOT.2024.3524255
[4] Park, H., Rafit I., and Mihaela, V., "Peer-to-peer networks: Protocols, cooperation and competition" in Streaming Media Architectures, Techniques, and Applications: Recent Advances. IGI Global, 2011. pp. 262-294, 2024.
[5] Behera, A., Champati, J., Morabito, R., Tarkoma, S. and Gross, J., "Towards Efficient Multi-LLM Inference: Characterization and Analysis of LLM Routing and Hierarchical Techniques." Journal of Latex Class Files, 14(8), pp. 06579-06595, 2025.
https://doi.org/10.48550/arXiv.2506.06579
[6] Rani S., Talwar R., Malhotra J., Ahmed, H., Sarkar, M., Song H. “A novel scheme for an energy efficient internet of things based on wireless sensor networks.” Sensors, 15(11), pp. 28603–28626, 2015.
[7] Li, P., Koyuncu, E. and Seferoglu, H. “Adaptive and resilient model-distributed inference in edge computing systems”, IEEE Open Journal of the Communications Society 4, pp. 1263-1273, 2023. doi: 10.1109/OJCOMS.2023.3280174
[8] Nandom, S., Abe, G. and Gambo, I. "Application of random forest and hierarchical clustering models for crop and fertilizer recommendation to farmers", Nigerian Journal of Technology, 44(1), pp. 114-122, 2025. https://doi.org/10.4314/njt.v44i1.13
[9] Sadiq, B., Zakariyya, O., Buhari, M., and Shuaibu, A., "Maximizing network capacity, control and management in designing a Telemedicine network: a review and recent challenges", Nigerian Journal of Technology, 43(1), pp. 80-100, 2024. https://doi.org/10.4314/njt.v43i1.11
[10] Rathod, V., Sable, N., Mali, Y., Ajalkar, D., Bharathi, M., and Padmaja, N., "A Network-Centred Optimization Technique for Operative Target Selection", Journal of Electrical Systems, 19(2), pp.87-96, 2023. https://doi.org/10.52783/jes.694
[11] Pang, R., Schroeder, H., Kynnedy, S., Barocas, S., and Xiao, Z., "Understanding the LLM-imitation of CHI: Unpacking the Impact of LLMs at CHI through a Systematic Literature Review." In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, pp.1-20, 2025. https://doi.org/10.1145/3706598.371372
[12] Rathod, V.,Gumaste, S., and Guttula, R., "Optimization of energy consumption in mobile Ad-Hoc networks with a swarm intelligence-based ABC algorithm." Discover Applied Sciences 7(8), p. 805, 2025. https://doi.org/10.1007/s42452-025-07472-6
[13] Wu, J., Yang, S., Zhan, R., Yuan, Y., Chao, L., and Wong, D., "A survey on LLMs-generated text detection: Necessity, methods, and future directions." Computational Linguistics, 51(1), pp. 275-338, 2025. doi: 10.1162/coli_a_00549
[14] Abujassar, R.S. “An Innovative Algorithm for Multipath Routing and Energy Efficiency in IoT Across Varied Network Topology Densities.” International Journal of Networked and Distributed Computing, 14 (1), pp. 1-20, 2025. https://doi.org/10.1007/s44227-024-00041-0
[15] Li, D., Jiang, B., Huang, L., Beigi, A., Zhao, C., Tan, Z., and Liu, H., "From generation to judgment: Opportunities and challenges of LLM-as-a-judge." In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, 2025 (1), pp. 2757-2791, 2025. doi:10.18653/v1/2025.emnlp-main.138
[16] Abbas, G., Ullah, S., Waqas, M., Abbas, Z., and Bilal, M., “A position-based reliable emergency message routing scheme for road safety in vanets.” Computer Networks, 213 (2022), p. 109097-109107, 2022. doi: 10.1016/j.comnet.2022.109097
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Nigerian Journal of Technology
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
The contents of the articles are the sole opinion of the author(s) and not of NIJOTECH.
NIJOTECH allows open access for distribution of the published articles in any media so long as whole (not part) of articles are distributed.
A copyright and statement of originality documents will need to be filled out clearly and signed prior to publication of an accepted article. The Copyright form can be downloaded from http://nijotech.com/downloads/COPYRIGHT%20FORM.pdf while the Statement of Originality is in http://nijotech.com/downloads/Statement%20of%20Originality.pdf
For articles that were developed from funded research, a clear acknowledgement of such support should be mentioned in the article with relevant references. Authors are expected to provide complete information on the sponsorship and intellectual property rights of the article together with all exceptions.
It is forbidden to publish the same research report in more than one journal.
How to Cite
