| 摘要: | 本論文研究使用現今(5G)第五代無線行動通訊網路中28G毫米波頻段適合做為室內涵蓋網路佈建主軸並利用機器學習(Fuzzy C-Means，簡稱FCM)模糊聚類以及(Particle Swarm Optimization，簡稱PSO)粒子群兩種不同的智慧演算法來模擬快速尋找出大型室內空間裡最適合的基地台位置點。
5G無線行動通訊基地台在佈建前開始都必須先進行現場環境勘查、確認佈署點周遭涵蓋環境、有無遮蔽物及空間大小高度，因此基地台的位置選擇就非常重要。(Millimeter Wave)毫米波頻段為實現5G行動通訊網路更大頻寬、更低延遲、更高速大量連結萬物的明日之星。其短波長、高路徑耗損、低穿透特性也是毫米波比起室外空間更適合大型室內涵蓋佈署。
研究利用功能強大的Matlab模擬軟體設計不同室內用途場景、不同的隨機人數用戶座標使用Fuzzy C-Means模糊聚類找出人群分佈質心點當作毫米波基站位置點部署策略；Fuzzy C-Means也常應用於通信網路場景中的節點部署藉由適當的質心數量將空間人群用戶分群達到基地台配置的效果。另一個PSO粒子群體智慧演算法也是廣泛出現在各種傳感器佈署的應用。
PSO 粒子群演算法利用群體能夠在沒有已知太多問題資訊或空間的情況下依據自身對環境的適應能力將群體中的個體引導往好的區域移動，有效率的藉由搜尋在欲求解的巨大空間問題內並找到適合解，將最後找到的解當作當作基地台的候選位置。研究目的在於利用Fuzzy C-Means和PSO兩種不同智慧演算法針對無線行動通訊網路的關鍵效能指標的影響其中包含用戶下載吞吐量、訊號加干擾雜訊比和總體覆蓋區域面積等.透過Matlab模擬產生毫米波基站位置，藉由模擬出的不同的基地台配置來了解兩種不相同的演算法在相同場景設計中的性能表現，加上對照以一般工程人員經驗做佈署其中的優劣差異能協助工程人員快速優化基地台的佈局，進而提升行動無線網路的整體效能和可靠性。
毫米波高頻通訊必須綜合考量到物理障礙、頻率繞射及穿透能力差、覆蓋面積不如低頻段通訊大等環境因素和終端用戶的移動行為條件。因此在目前現實環境電信業者將毫米波網路建置於大型室內空間並盡量以最大面積及最密集用戶做涵蓋佈局。其次，室內環境因素如不同材質的隔間、裝潢等其它電氣設備存在皆可能對信號品質造成干擾， 工程人員透過Fuzzy C-Means及PS0智慧粒子群兩種演算法的比較在基站建設前透過既有建築空間尺寸大小及人流密集度快速找出更適合安排毫米波基地台點的配置，觀察配置點的無線訊號品質、強度以確保最大面積的無線覆蓋、一致性和無線網路連接的穩定性、提高行動無線網路的效能和可靠性，為用戶帶來更優質的網路體驗則是本研究重點。
;This thesis investigates the deployment of 28GHz millimeter wave (mmWave) technology, suitable for indoor coverage in 5G mobile communication networks. The research employs two different intelligent algorithms—Fuzzy C-Means (FCM) clustering and Particle Swarm Optimization (PSO)—to rapidly identify the optimal locations for base stations in large indoor spaces.
Before deploying a 5G mobile communication base station, site surveys are necessary to assess the environment, including coverage areas, obstacles, and spatial dimensions. Hence, the choice of base station location is crucial. Millimeter waves, characterized by short wavelengths, high path loss, and low penetration, are more suited for indoor environments compared to outdoor spaces. These properties also enable 5G networks to achieve higher bandwidth, lower latency, and enhanced connectivity.
This research utilizes the powerful simulation software Matlab to design various indoor scenarios and generate random user coordinates. Fuzzy C-Means is used to identify the centroids of user clusters, which serve as potential deployment points for mmWave base stations. FCM is frequently applied in communication network scenarios to deploy nodes by clustering user populations around optimal centroids. PSO, another widely used intelligent algorithm, is common in sensor deployment applications. It guides particles in a population towards better regions based on their adaptability to the environment, efficiently searching through large solution spaces to find optimal base station locations.
The goal of this study is to evaluate the impact of these algorithms on key performance indicators of wireless communication networks, including user throughput, Signal to Interference plus Noise Ratio (SINR), and overall coverage area. By simulating mmWave base station deployments in Matlab, the performance of FCM and PSO under identical scenario designs is compared. Additionally, the study contrasts these algorithmic results with the traditional deployment strategies used by engineers to identify their advantages and disadvantages, helping to optimize base station layouts quickly and enhance the overall performance and reliability of mobile networks.
Due to the high frequency of mmWave communication, several factors, including physical obstacles, diffraction, poor penetration capabilities, and reduced coverage area compared to lower frequency bands, must be considered along with the mobility of end users. Currently, telecommunications operators focus on deploying mmWave networks in large indoor spaces, aiming to maximize coverage and user density. Indoor environmental factors, such as partition materials, decor, and electrical equipment, can also interfere with signal quality. By comparing FCM and PSO, this study assists engineers in quickly identifying better mmWave base station locations based on existing building dimensions and user density. The focus is on optimizing signal quality and strength at deployment points to ensure maximum coverage, network consistency, and stability, thereby improving the overall performance and reliability of mobile networks for users. |
