2014-08-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/711215摘要：隨著使用者數量的增加，新世代無線網路將會變得前所未有地複雜：不單是干擾問題會限制整體的系統效能，離散的各節點之間未整合的行為將使系統更加動態化。而近年來由於在網路消息理論(network information theory)中的突破，對於干擾在網路通訊中的影響有了本質上更清楚的理解，從而有了更先進的干擾管理技術。在這當中，節點間的互動(interaction)如回饋(feedback)、合作(cooperation)等，由於理論上能乘數倍地提升系統容量，而成為最被看好能夠突破現今因干擾而樹立的頻譜效率限制的技術競逐者。然而，在消息理論的相關文獻中，大部份均假設網路中的節點連接方式為固定的，而節點之間的高斯通道可能是常數通道或衰變通道。此一假設在無線數據網路中的實體層並非永遠成立，原因在於各接取點的媒體存取控制層(MAC layer)機制和網路層(network layer)協定是分散式的，而造成節點之間的通道可能是間歇性地連接。一方面倘若利用干擾的間歇性，則系統的傳輸容量原則上可能可以進一步提升；另一方面，若節點之間的互動也具有間歇性的特質，則互動式干擾管理的技術可能也會受影響而需要改進。本計畫提出的創新中最重要的是，我們發現當干擾是間歇性時，一個bit的信號回饋可以提供無上限的通道容量(capacity)增益。 本計畫的主要目標是要探討，在具備間歇特性的無線網路之中，節點互動在干擾管理上扮演的基礎角色，特別是當干擾是間歇時或互動是間歇時如何使用節點互動來作干擾管理。同時，本計畫也將發展有效率的編碼方式及系統設計法則，以期能在未來的無線系統(如5G行動通訊)中使用。<br> Abstract: Modern wireless networks are complex – as the number of users grows, not only does interference limit the overall performance, but the decentralized behaviors of distributed terminals result in highly dynamical systems. Recent advances in network information theory lead to better fundamental understanding about interference, and, consequently, more advanced schemes for interference management. Among these schemes, distributed terminal interaction, such as cooperation and feedback, emerges as one of the strongest contenders, because it can boost up system capacity multiplicatively. However, in the theoretical modeling at hand, it is usually assumed that network topology is fixed and associated with constant channel gains or fading channel coefficients. Owing to the intermittence and burstiness caused by the decentralized medium-access-control mechanism and/or networking protocols, this assumption may be far off in wireless data/ad-hoc networks, especially from the perspective of the physical layer. Such intermittence and burstiness can either yield significant gain in system capacity, if one exploits the fact that interference is bursty, or degrade the performance of the theoretically devised interactive schemes, if the resource or opportunity for interaction is intermittently available. In particular, a key innovation in this proposal is that, when interference is intermittent, one-bit feedback provides unbounded capacity increase. The goal of this project is to explore the fundamental role of interaction in bursty wireless networks through the lens of information theory, especially in how interaction helps manage interference when interference and/or resource for interaction is intermittent. We also aim to devise efficient coding schemes and develop system design insights for future wireless networks, such as 5G cellular systems.