DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Cho, Dong-Ho | - |
dc.contributor.advisor | 조동호 | - |
dc.contributor.author | Chung, Byung Chang | - |
dc.contributor.author | 정병창 | - |
dc.date.accessioned | 2018-05-23T19:37:47Z | - |
dc.date.available | 2018-05-23T19:37:47Z | - |
dc.date.issued | 2017 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=675845&flag=dissertation | en_US |
dc.identifier.uri | http://hdl.handle.net/10203/242040 | - |
dc.description | 학위논문(박사) - 한국과학기술원 : 전기및전자공학부, 2017.2,[iv, 61 p. :] | - |
dc.description.abstract | In this thesis, we aim to improve the performance of the coexisting networks based on game theory. Coexisting networks refer to the network which consists of multiple radio access technologies (RATs) sharing the same frequency band. Such a coexisting network, inter-RAT interference is the most major factor for degrading the throughput performance. Thus, in this thesis, we apply game theoretic approaches to reduce the inter-RAT interference. However, it is hard to formulate non-cooperative game since the gain and loss of players are vague. On the other hand, in cooperative game model, we can reach the better equilibrium point through bargaining or collaborative behavior. In the first part, we propose the transaction among macrocell and APs with the use of data offloading and almost blank subframe (ABS). Each AP admits the offloaded data from macrocell and obtains more ABS. With the proposed scheme, each player can serve its subscribers who has better channel condition. We verify the 45 % gain of proposed algorithm by means of a simulation. In the second part, we suggest bargaining among location-based groups when a macrocell uses beamforming technique. If the macrocell can make the beamforming pattern, ABS can be applied spatially. APs that are not in the beam direction can support their receivers with low interference level. Therefore, we first group the users according to their locations and formulate bargaining game among groups. With ABS and beamforming, we can notice that the throughput is improved by 40 %, compared to conventional schemes. In the third part, we proposed network operator selection based on reinforcement learning. If there are multiple network operator, APs have to determine appropriate network operator to cooperate with. From the selected network operator, APs can receive reserved resource of network operator. Thus, APs should know the traffic information of network operators. To collect traffic information, we adopt reinforcement learning algorithm. Using reinforcement learning, APs adaptively change network operator. From the simulation, the throughput is improved by 20 %. Above three schemes are based on game theory framework. Since we utilized game theory, proposed algorithms could consider the incentives of every player. Moreover, proposed algorithm runs with low complexity compared to optimal algorithm. Thus, we expect that our proposed algorithm can be widely used in coexisting communication networks. | - |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | coexisting network | - |
dc.subject | cooperative game | - |
dc.subject | data offloading | - |
dc.subject | coalition game | - |
dc.subject | bargaining game | - |
dc.subject | reinforcement learning | - |
dc.subject | 공존 통신망 | - |
dc.subject | 협조적 게임 | - |
dc.subject | 데이터 오프로딩 | - |
dc.subject | 연합 게임 | - |
dc.subject | 협상 게임 | - |
dc.subject | 강화 학습 | - |
dc.title | Intellectual Algorithm for Enhancing System Capacity in Cellular and WLAN Coexisting Communication Networks | - |
dc.title.alternative | 셀룰러와 무선 LAN이 공존하는 통신망에서 시스템 용량을 개선하기 위한 지능 알고리즘 | - |
dc.type | Thesis(Ph.D) | - |
dc.identifier.CNRN | 325007 | - |
dc.description.department | 한국과학기술원 :전기및전자공학부, | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.