電機資訊學院: 電信工程學研究所指導教授: 張時中; 魏學文馬琳茵Ma, Lin-YinLin-YinMa2017-03-062018-07-052017-03-062018-07-052016http://ntur.lib.ntu.edu.tw//handle/246246/276235隨著無線網路流量以及應用服務大量的成長，預期電信營運商在人口擁擠區需要佈建大量的小型基站設施來達到資料容量和傳輸速率需求，這也讓電信營運商的資本性和營運性支出大幅提升。鑑於當前基站接取分享(RAN Sharing) 及無線網路虛擬化(Wireless Network Virtualization)技術的發展，為電信營運商帶來LTE小型基站與頻譜分享市場的可能性，稱之為MoSCS (Multi-Operator LTE Small Cell Sharing Market)，不僅可減少重複佈設，也可節省相當的佈建成本並提升總營收。在MoSCS市場中，營運商兼具兩種角色:資源分享者(Providing Operator, PO)與資源需求者(Accessing Operator, AO)。營運商身為PO，賣自己的資源給自己的客戶和其他的營運商(AO)，也為了滿足自己客戶的需求，扮演AO的角色，向其他營運商(PO)尋求資源。由於營運商(PO)間彼此互為競爭對手，自然不願分享其資源與資訊，是MoSCS市場分享合作構想的基本挑戰。 有鑑於公共服務業競爭規管上，透過一個受政府監管、公正且獨立的第三方來進行寡頭競爭者間的協調，是一種具成功案例的模式，本論文研究據以提出一創新的獨立服務分享協調者(Independent Service Sharing Coordination- ISSC)架構設計構想。基本設計問題包括下列幾項: Q1: 如何設計一個獨立服務分享協調機制架構，在不透露營運商彼此間的競爭資訊要求下，來促成營運商間願意合作分享彼此的基站與頻譜資源? Q2: 在ISSC架構下如何根據營運商的需求，設計協調分享頻譜與基站資源運用的機制? Q3: ISSC要如何訂定合理的接取服務價格，讓營運商可透過分享資源提升各自的營收? 針對以上問題，本論文研究為基站與頻譜資源分享市場設計了創新的ISSC分享協調架構與配套機制： 1. 創新的協調架構設計： 此架構為新構想，由資源預測、資源分配、價格協調與用戶預測等部分所構成。其中ISSC平台收集各營運商之必要資訊並予保密，再以各營運商水平競爭的模式計算營運商間資源與價格之均衡，據以協調各營運商，創造共贏之誘因。 2. 疊代式價格競爭與資源配置合作ISSC協調機制設計： ISSC協調機制考量營運商兼具賣家(PO)與買家(AO)的角色，結合了Bertrand賽局與Bankruptcy賽局來設計。 (2.1)ISSC以Bertrand價格競爭賽局估算各區域PO競爭之均衡價格與相對應AO需求量 在一區域當中，各營運商(PO)於供給端彼此之間為寡頭競爭市場。本論文研究將資源限制先視為軟性限制，利用PO間價格競爭來估計各AO分享需求, 因此將此問題形成為Bertrand賽局求解Nash均衡接取服務價格及估計相對應的各AO需求量。由ISSC求解能減少營運商間自行反覆議價的大量資訊傳輸與時間。 (2.2)當AO之需求大於PO基站之供給時，ISSC依據資源限制及各AO需求量以Bankruptcy賽局按需求貢獻度公平分配給AO分享所需資源 在一PO分享之基站上，因AO資源限制為軟性限制，若PO資源不足提供給各AO，本論文將此問題類比為合作式Bankruptcy賽局，並在ISSC平台中根據各AO對於資源短缺的貢獻度，以Shapley value的計算方式來分配PO之基站有限資源給AO。 (2.3) ISSC 以用數戶移動與用戶需求模型疊代求取價格與資源需求 ISSC可經由預測營運商需求和Shapley資源分配差異的結果，首先，ISSC得到營運商的用戶滿意程度，在Poisson-based用戶轉移模型與用戶需求模型中疊代反覆預測總用戶數與用戶需求，最後進一步分析用戶數變化/用戶需求對營運商之需求、價格、總營收的影響。 本論文根據用戶分佈設計不同的場景來進行所設計機制的數值驗證與分析。結果顯示，總用戶數、價格、頻譜與基站資源配置在提出之場景中，ISSC演算法疊代結果皆會趨於收斂。因在用戶轉移模型中，若營運商的用戶滿意度較差，將以Poisson機率流失其用戶，在此同時，他方營運商將增加用戶。用戶減少之營運商，因用戶平均可得較多資源，其用戶滿意度持續提升；而用戶提升之營運商滿意度漸而下降。當營運商間滿意度隨著用戶變化來到平衡，系統趨於穩定。而在用戶需求模型當中，營運商接取頻譜後提供的服務品質提升，用戶需求將隨之增加但不轉移至其他營運商。因營運商分配的到的資源有限，隨著用戶需求增加，服務品質下降，使得用戶需求漸而減少。當用戶需求達到平衡時，ISSC演算法疊代結果趨於穩定。 比較各場景中營運商的總營收增長幅度及總用戶數間之關係的結果顯示，若各基站皆有足夠資源可互相分享(Low density scenario)，相較於無ISSC協調、各自營運而不分享，基站使用率可提升至近100%，各營運商之總營收增加3~6倍且不互相競爭其用戶，為一雙贏局面。而在基站資源不足以提供給需求者(High density scenario)情境下，各營運商總營收仍能增加，服務品質較差之營運商將損失一定程度之客戶，直至服務品質和其他營運商相近。營運商雖損失其客戶，但透過ISSC提供之AO對各基站需求與PO分配的結果，營運商可評估是否在該不被滿足區域佈建基站，以滿足該區客戶需求，進而提升服務品質，避免流失客戶。 本論文使用AppServ架站並結合Python實作LTE小型基站分享的線上獨立服務分享協調者平台(Online ISSC Platform)。AppServ包含Apache, PHP和 MysSQL等網站環境元件，可在Windows作業系統中執行Window-Apache-MySQL-PHP (WAMP) 解法堆疊並允許透過phpMyAdmin來管理MySQL資料庫。本論文利用Python 來執行內部演算法並顯示於平台中。With the tremendous growth in wireless data traffic and services, mobile network operators (MNOs) may need to deploy large amount of network infrastructure to meet the unprecedented capacity and data rate demand, which increases capital expenditures (CAPEX) and operating expenses (OPEX) correspondingly. The development of radio access network sharing (RAN sharing) and wireless network virtualization (WNV) brings the possibility of small cell sharing. Each MNO can reduce large amount of de-ploying cost and gain profits by cooperation. In this thesis, we define multi-operator LTE small cell sharing market (MoSCS) as multiple operators share their LTE small cells and spectrum resources. In this market, the MNO plays two roles simultaneously: Providing Operator(PO) and Accessing Operator(AO). As a PO, MNO sells small cell resources to its own subscribers and other AOs. To satisfied its own subscriber’s de-mand, the MNO plays as an AO, seek for resources from other POs. However, the MNOs are competitors to each other, which brings small cell sharing market many problems. Especially, the MNO may not want to share their resource and information to the competitors. A fair and independent third-party coordinator, supervised by government, is a success model in the public service competition. Therefore, in this thesis, we propose a novel coordination framework called independent service sharing coordination (ISSC). The basic design problems include: Q1. How to design a ISSC framework to facilitate the MNO to share its small cell and spectrum resources without revealing MNO’s information? Q2. Under the ISSC framework, how to design the small cell resource allocation mechanism based on the demand requirement of different MNOs? Q3. Under the ISSC framework, how to determine a reasonable access price to enhance the profit of each MNO by sharing resources? Based on the above problems, we proposed a novel ISSC framework and the cor-responding mechanism design for MoSCS market in this thesis. (1) Designed a novel coordination framework: This framework is a novel idea, which consists of resource estimation, resource allocation, price determination and subscriber estimation. ISSC platform ensures the security of the information provided by MNOs. Then, ISSC calculates the equilibrium of price and resource demand to create a win-win situation for MNOs. (2) Designed an iterative ISSC mechanism with competitive pricing and coopera-tive resource allocation: The ISSC mechanism design combines Bertrand game and Bankruptcy by consid-ering two roles of MNO: PO and AO. (2.1) ISSC use Bertrand game to model the equilibrium of competitive price of PO to the corresponding demand of AO In an area, each PO compete with each other, which is a monopoly market. In this thesis, we first model the demand restriction as soft restriction. Then, by using the competitive pricing to estimate the demand requirement of each AO. Therefore, we formulate this problem into Bertrand game and find Nash equilibrium price and the corresponding demand of AO. Solving the Nash problem by ISSC can reduce large amount of information transmission and negotiate time. (2.2) When the demand of AOs is large than a PO’s small cell can supply, ISSC allo-cates the resources based on resource restriction and the contribution of demand re-quirement of an AO by using Bankruptcy game In this thesis, we formulate the allocation problem into Bankruptcy game when the PO’s small cell resources can not satisfy the demand of AOs. Moreover, we use Shapley value to allocate PO’s resources to AO in the ISSC platform, which is an algo-rithm allocates resources based on the contribution to the shortage of resources. (2.3) ISSC use subscriber to find the price and demand requirement iteratively ISSC estimate the quality of each MNO by the difference between AO’s demand and the resource allocation result by using Shapley. Then, by using a Poisson-based subscriber transition model to estimate subscriber number iteratively, we can further analyze the effect of subscriber number to the demand, price and total payoff. In this thesis, we verify and analyze our mechanism by designing several scenari-os based on different subscriber density of MNO in the MoSCS market. The simulation results show that the total subscriber number, price and resource allocation of small cell will converge in our proposed scenarios. Because in our subscriber estimation model, if the MNO’s quality is poor, its will lose its subscribers in a Poisson distribu-tion. Meanwhile, its competitors’ subscriber will increase. As a MNO’s subscriber de-crease, the remain subscriber can get more resource. Therefore, the quality of the MNO increase. In other words, the MNO’s quality will decrease as its subscribers increase. When the quality between MNO become balance, the system will tend to stable. By comparing the results of total payoff and the subscriber number, if the small cells resources are enough (Low density scenario), the utilization of base station can raise up to nearly 100%. Each MNO’s total payoff increase to 3~6 times and do not compete other MNO’s subscribers, which is a win-win situation. When the small cells can not provide enough resources to AOs (High density scenario), each MNO still gain profit. However, the MNO with poor quality will keep losing of its subscribers not until its quality is raised. Though MNO lose its subscribers, but through the estimated results of AO to the demand of POs from ISSC platform, MNO can evaluate whether to deploy new small cells. The deployment may satisfy the requirements of subscribers and im-prove the quality, which will avoid the subscriber to leave. We implement an online ISSC platform for LTE small cell sharing market by us-ing AppServ and Python. AppServ includes Apache, PHP and MySQL, which allows window system to run solution stack called Windows-Apache-MySQL-PHP(WAMP) and manage MySQL database through phpMyAdmin. We use python scripts to run our algorithm and update the results to the ISSC platform.7273308 bytesapplication/pdf論文公開時間: 2019/10/5論文使用權限: 同意有償授權(權利金給回饋本人)營運商間LTE小型基站分享市場線上獨立服務分享協調者平台資源分配奈許均衡價格破產賽局伯川德競爭模型行動接取網路共享Multi-operator LTE small cell sharing marketonline ISSC platformresource allocationNash equilibrium-based pricebankruptcy gameBertrand competition modelRAN sharingthesis10.6342/NTU201603075http://ntur.lib.ntu.edu.tw/bitstream/246246/276235/1/ntu-105-R03942110-1.pdf