Robust Average Consensus under Byzantine Attacks
We study the problem of average consensus in multi-agent systems where some of the agents may malfunction. The object of robust average consensus is for non-faulty agents to converge to the average value of their initial values despite the erroneous effects from adversarial agents. To this end, we p...
Uloženo v:
| Vydáno v: | Proceedings of the IEEE Conference on Decision & Control s. 3839 - 3844 |
|---|---|
| Hlavní autoři: | , , |
| Médium: | Konferenční příspěvek |
| Jazyk: | angličtina |
| Vydáno: |
IEEE
16.12.2024
|
| Témata: | |
| ISSN: | 2576-2370 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | We study the problem of average consensus in multi-agent systems where some of the agents may malfunction. The object of robust average consensus is for non-faulty agents to converge to the average value of their initial values despite the erroneous effects from adversarial agents. To this end, we propose a surplus-based consensus algorithm that can achieve robust average consensus under Byzantine attacks in the multi-agent networks with directed topologies. The key idea is to equip each normal agent with a running-sum variable so that it can record the effects from/to neighbors across iterations. Moreover, compared to the existing secure broadcast and retrieval approach where each agent keeps track of the initial values of all agents in the network, our algorithm saves massive storage especially for large-scale networks as each agent only requires the values and the correct detection of neighbors. Finally, numerical examples are given for verifying the effectiveness of our algorithm. |
|---|---|
| ISSN: | 2576-2370 |
| DOI: | 10.1109/CDC56724.2024.10886555 |