動態分時多工長期演進技術之具體驗品質感知細胞間干擾抑制

Abstract

在動態分時雙工(Dynamic TDD)系統中，為了依據實際上下行需求進行較佳資源配置，而支援7個上下行配置(uplink-downlink configuration)，如此藉由在時間軸上不同的上下行配置達到上下行非對稱資源配置。然而在密度高的小型基地台佈建的網路中，由於基地台的地理位置鄰近，加上每個基地台各自設定其上下行配置，彼此會產生嚴重的干擾，鄰近的兩個基地台皆為上行或皆為下行就會有同向傳輸干擾；一者為上行另一者為下行則會有異向傳輸干擾。本文針對此一系統提出一個干擾抑制的方法，依據使用者的體驗品質來最佳化分配基地台的資源，一方面降低基地台之間的干擾，另一方面維持系統整體的使用體驗品質。模擬結果顯示我們所提出的方法，不論在同質網路與異質網路，都能讓系統有更高的且更穩定的體驗品質。最後我們也基於模擬結果提出一些關於同向及異向傳輸干擾與使用者經驗品質管理的洞見，以利日後設計或運行Dynamic TDD的小型基地台網路。

In LTE dynamic time division duplexing (dynamic TDD), seven uplink-downlink configurations (UL-DL configurations) are supported to accommodate downlink/uplink traffic asymmetry for traffic adaptation. The Third Generation Partnership Project (3GPP) has considered the opportunity of adopting such dynamic adaptation of UL-DL configurations in small cell networks to enlarge the network capacity. However, dynamic TDD inducing opposite transmission directions in different cells results in new destructive interference components, i.e., BS-to-BS and UE-to-UE interference, which makes the inter-cell interference in small cell networks even worse. In this paper, a QoE-aware interference mitigation scheme is proposed to mitigate interference and to manage QoE as well in small cell deployment networks. The scheme explicitly uses the received signal strength and QoE state of users to generate a QoE-aware interference graph, where vertices represent cells, edges represent intolerant inter-cell interference among the incident cells and weights represent the utility increment of QoE. Then, a graph-based optimisation problem is formulated to determine which cells are allowed to transmit to maximize the system utility of QoE. Further, a linear-time approximation algorithm is proposed for large-scale deployment environments. The LTE-based simulation results show that the proposed QoE-aware interference mitigation scheme significantly improves overall QoE and also alleviate inter-cell interference. Finally, based on the simulation, we provide some insight into QoE management and interference mitigation in both homogeneous and heterogeneous networks operating dynamic TDD.