Researches
AI based multi-path selection
Multi-media multi-path(MMMP) system is a system in which utilizes various networks simultaneously. It is expected that MMMP system can enhance communication speed, reliability, security of network. In this research, we focus on intelligent network control in MMMP system. Recurrent neural network(RNN) is applied to predict the network performance, according to the previous time-series network performance parameters. Moreover, we implemented testbed learning server to evaluate the performance of proposed algorithm.
The exhibition & tests are on
- World IT Show 2018
- Mobile World Congress 2019 (Barcelona, Spain)
- DX Korea 2020
- KOREN Seoul-Pangyo-Daejeon and Seoul-Suwon-Daejeon paths (미래네트워크선도시험망 서울-판교-대전, 서울-수원-대전 구간)
- Military networks including wired, microwave and satellite network (실제 국방망)
Cascade AI structure for detecting network intrusion
Artificial intelligence(AI) is applied to improve accuracy in network intrusion. AI is a great optimization method, but the problem of lack of data often arises. To solve this problem, we suggest cascade AI structure to solve data labeling problem. By applying unsupervised learning, we can reduce the difficulty of data labelling.
Next-generation communication systems
Future networks will become more complex depending on user/service requirements. Delicate optimization is required to meet these requirements. In order to reduce the complexity of existing optimization, we are interested in multi-agent learning, neural networks and game-theoretic modeling.