TY  - JOUR
AU  - Yogi, Manas Kumar
AU  - Kavya, Majji
PY  - 2026
DA  - 2026/06/17
TI  - Scalable Trust through Strategic Verification: A Game-Theoretic Framework for Multi-Agent Systems
JO  - Next-Generation Computing Systems and Technologies
T2  - Next-Generation Computing Systems and Technologies
JF  - Next-Generation Computing Systems and Technologies
VL  - 2
IS  - 2
SP  - 35
EP  - 50
DO  - 10.62762/NGCST.2026.430053
UR  - https://www.icck.org/article/abs/NGCST.2026.430053
KW  - multi-agent systems
KW  - game-theoretic verification
KW  - byzantine fault tolerance
KW  - federated learning security
AB  - These days, many eco-systems related to federated learning, blockchain, self-driving cars, and scientific computing have many agents working together, each doing its own part. Using a single central system to check if all the agents are doing their work correctly is slow and gets more expensive as more agents are added. This paper introduces a new way called the Verification Game (VG). The agents don’t depend on a central system. The agents check each other’s work. If the agents are honest, they get rewards, so telling the truth is the best option. This type of method also saves a lot of computing power because it doesn’t check every single task. We also came up with a method called Adaptive Strategic Verification (ASV) that figures out which agent’s work should be checked. It saves the computing resources because only selected tasks are verified, not everything. From the year 2020 to 2025 we tested six real-world systems. These included things like learning systems where the data is shared across devices, groups of self-driving cars, blockchain networks, big computer clusters, supply chains, and websites that monitor the content. On average, the system could spot the problems up to 97.3% of the time and only used about a quarter of the computing power needed by central systems. Agents learned to be honest within 31–48 rounds (or equivalent time, mean approximately 40 rounds) depending on domain. Even when most agents tried to cheat together, the system was able to stop collusion. Overall, this approach helps to lots of agents that work together fairly without depending on a central system. It makes building large and reliable systems easier.
SN  - 3070-3328
PB  - Institute of Central Computation and Knowledge
LA  - English
ER  - 

@article{Yogi2026Scalable,
  author = {Manas Kumar Yogi and Majji Kavya},
  title = {Scalable Trust through Strategic Verification: A Game-Theoretic Framework for Multi-Agent Systems},
  journal = {Next-Generation Computing Systems and Technologies},
  year = {2026},
  volume = {2},
  number = {2},
  pages = {35-50},
  doi = {10.62762/NGCST.2026.430053},
  url = {https://www.icck.org/article/abs/NGCST.2026.430053},
  abstract = {These days, many eco-systems related to federated learning, blockchain, self-driving cars, and scientific computing have many agents working together, each doing its own part. Using a single central system to check if all the agents are doing their work correctly is slow and gets more expensive as more agents are added. This paper introduces a new way called the Verification Game (VG). The agents don’t depend on a central system. The agents check each other’s work. If the agents are honest, they get rewards, so telling the truth is the best option. This type of method also saves a lot of computing power because it doesn’t check every single task. We also came up with a method called Adaptive Strategic Verification (ASV) that figures out which agent’s work should be checked. It saves the computing resources because only selected tasks are verified, not everything. From the year 2020 to 2025 we tested six real-world systems. These included things like learning systems where the data is shared across devices, groups of self-driving cars, blockchain networks, big computer clusters, supply chains, and websites that monitor the content. On average, the system could spot the problems up to 97.3\% of the time and only used about a quarter of the computing power needed by central systems. Agents learned to be honest within 31–48 rounds (or equivalent time, mean approximately 40 rounds) depending on domain. Even when most agents tried to cheat together, the system was able to stop collusion. Overall, this approach helps to lots of agents that work together fairly without depending on a central system. It makes building large and reliable systems easier.},
  keywords = {multi-agent systems, game-theoretic verification, byzantine fault tolerance, federated learning security},
  issn = {3070-3328},
  publisher = {Institute of Central Computation and Knowledge}
}