2016-08-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/713402摘要：物聯網 (IoT) 在下世代無線通訊系統中將會扮演重要的角色。一方面,IoT 將會創造許多無線通訊 產業中的機會,包括了對穿戴式可連網裝置的需求量提升,以及產生各種新式應用的可能性等。另一 方面,為了滿足這些需求,在 IoT 裡的連網功能以及資料處理中,有很多基礎理論上的挑戰,而這些 挑戰與傳統無線通訊系統中所遇到的截然不同。在連網功能方面,這些獨特的挑戰包括:間歇性的資 料傳輸需求、異質的延遲與能量限制、巨量非同步以及未協調的接取等。在資料處理與分析方面,則 有分散式資料處理與分析、大範圍蒐集的不可靠資料、隱私考量等挑戰。 本計畫中,透過分析資訊傳輸、壓縮、運算以及資料分析在 IoT 限制條件下的基本理論極限,我 們將探索並嘗試解決以上所述 IoT 帶來的挑戰。在連網功能方面,我們的目標是發展關於 IoT 的新實 體層理論,進而提出關於建置新系統的核心概念以及新的編碼方法。在資料處理與分析方面,我們的 目標是發展在通訊限制下的統計推論與計算等的理論表現值,同時納入隱私限制的考量。值得注意的 是,在我們的初步成果中,我們已經發現在大型無線隨意網路中,若納入傳輸時間的限制,則整體傳 輸速率對於用戶數目的縮放律 (scaling law) 將會取決於傳輸時間對於用戶數目的縮放律。此初步成果 顯示,在 IoT 的限制之下,無線網路的表現可能與傳統網路有很大的差異,值得進一步探索,以發展 更佳的 IoT。<br> Abstract: Internet of Things (IoT) is expected to play a key role in the next-generation wireless communication systems. On one hand, IoT creates many great opportunities for wireless industry, including promising growth in the demand of wearable portable devices, and opens up immerse possibilities for new and fascinating applications. On the other hand, it is network connectivity and data processing that are critical to fulfill all these promises about IoT, and both of them address unique challenges that are critically different from traditional wireless communication systems. For network connectivity, the unique challenges include (1) bursty and sporadic data, (2) heterogeneous latency and energy constraints, and (3) massive asynchronous access. For data processing, the unique challenges include (1) distributed data processing, (2) large-scale noisy data, and (3) privacy-utility-communication trade-off. In this project, we aim to address the above challenges by exploring and investigating the fundamental limits on information flow and data analysis in IoT networks. For the connectivity part, our goal is to develop a new fundamental physical layer (PHY) theory for IoT, so that new insights regarding system design and new coding schemes can be devised. For the data processing part, our goal is to identify the performance bounds in statistical inference and function computation from large-scale IoT networks, with or without privacy considerations. Our preliminary results show that, the asymptotic scaling law of the total throughput of an ad hoc network depends critically on what is the scaling regime of the overall latency constraint. This implies that the fundamental performance of IoT networks may be critically different from the traditional ones, and hence it is worthy to investigate the fundamental limits of information and data under IoT-kinds of constraints so that a better IoT network can be built based upon the insights developed in these theoretical findings.物聯網間歇非同步接取資料隱私Internet of ThingsNon-Asymptotic Information TheoryAsynchronous and Bursty CommunicationData PrivacyDifferential Privacy