可自我調適之ZigBee 網路與感測器應用

Abstract

近年來，許多的無線感測網路系統採用ZigBee當成它們的通訊協定。ZigBee是由ZigBee聯盟為了低成本、低資料傳輸率、以及低電量的無線感測網路所提出的標準。它採用IEEE 802.15.4的實體層及媒體存取控制層，其後延伸至網路、應用、以及一些安全服務等。由於其自我調適的特性，無線感測網路已經廣泛的應用在一些監控的應用上。在這論文中，我們探討了一些自我調適的ZigBee網路中的議題，以及感測器應用。

在第一個研究中，我們探討ZigBee的定址問題，我們提出兩個自動參數的選擇機制，來輔助ZigBee網路的定址，這兩個自動參數選擇機制，可事先探測網路的資訊，可以幫助ZigBee中樹狀網路的定址，減少孤兒的問題。而第二個研究中，我們另外提出一個允許調借位址之分散式位址分配演算法，這個演算法允許一個父節點去跟它的鄰居節點調借一整棵子樹的位址空間以減輕孤兒問題。當一個新的節點試著連結上一個父節點路由器，而此父節點並無任何多餘的位址空間時，這個父節點將會去跟它的兩跳內的鄰居去調借一整棵子樹的位址空間。我們亦提出一個輕量級的路由方式以支援一對一的路由。

在第三個研究中，我們對於物聯網內的多播協定有興趣。近來許多物聯網或無線感測網路的應用會採用多播的方式當成它們的通訊協定：然而由於ZigBee所提供的多播協定會造成非常高的封包開銷及感測器節點能量的耗損。於是我們提出一個可相容於ZigBee網路的能源節省多播協定，藉由我們的多播機制，每一個感測器節點會使用維持模組來維持它的鄰居集群組成員資訊：而在多播時，感測器節點會使用多播模組來競爭成為轉信者；另外，每個感測器節點會偷聽它鄰居的廣播訊息用以避免不必要多餘的傳輸以及確保自己傳送的訊息有成功的被接收。

第四個研究中，我們提出一個微氣候監控的場景，利用車載感測器網路，各個車輛載具上裝設著一些感測器來達到精細的監測。而隨著車輛的移動，這輛車可以在不同的位置收集到一些的感測結果。我們提出一個車載網路的架構來收集及監測都市內的空氣品質，而其中的車輛的行進是不可控制的，如計程車。而在此架構下，衍伸出了兩個議題，(1)如何調適這些行動車輛的回報頻率，在最少通訊成本的情況下卻能達到一個使用者所需求的監控品質。(2)如何利用機會式的通訊來減少訊息的傳輸量。我們提出了演算法來解決這兩個議題並透過模擬來驗證它們的效能。另外，我們也發展一套基於ZigBee的原型系統來監控都市內的二氧化碳濃度。

Recently, many WSN systems have adopted ZigBee as their communication protocol. ZigBee is designed for low cost, low data-rate, and low-power WSN by the ZigBee Alliance. It adopts the physical (PHY) and the medium access control (MAC) layers defined by IEEE 802.15.4 and extends to network, application, and security services.

Due to self-adaptable characteristic, WSNs have been widely adopted in surveillance and monitoring applications. In this dissertation, we probe some issues and application in self-adaptable ZigBee networks. In the first work, we are interested in addressing issue in WSNs. Addressing in WSNs is to assign each newly-joining device a unique address. However, to allot the naming space in a large-scale distributed WSNs is not an easy task. ZigBee is a popular communication standard for wireless sensor networks. It suggests a distributed address assignment mechanism. A parent device can calculate network addresses for its child devices without communicating with other devices. However, the parameter configuration of this mechanism strictly restricts the number of children of a device and the depth of the network. ZigBee does not recommend suitable parameters. The improper parameter configuration usually makes many devices isolated from the network which become orphan devices. Two automatic parameter selection schemes are proposed for ZigBee address assignment by probing the network and then selecting parameters in advance to alleviate the orphan problem. They can automatically suggest proper parameters for different network topologies and thus help the original ZigBee address assignment mechanism to effectively reduce orphans. In the second work, we propose a distributed address assignment scheme by allowing a parent to borrow a subtree of address space from a neighbor to alleviate the orphan problem. When a new node tries to associate with a parent router which has no free address, this parent router will inquire its 2-hop neighbors for lending a subtree of free address space. We also propose a light-weight routing to support one-to-one routing in this environment.

In the third work, we are interested in the multicast protocol in Internet of Things (IoT). Recently, many Iots or WSN applications adopt ZigBee as their communication protocol. In these applications, messages may need to be disseminated to some specific objects or nodes using multicast transmissions. However, we observe that the original ZigBee multicast protocol causes extremely high packet overhead and energy consumption. In this work, we propose a ZigBee-compatible energy efficient multicast protocol, which allows nodes to execute the designed procedures in a distributed manner. By our scheme, each node uses the designed maintenance module to manage its neighbors. When multicasting, a node utilizes the designed multicasting module to compete for being a relay node by the proposed backoff mechanism. In addition, each node overhears its neighbors' broadcasts to avoid unnecessary transmissions and to ensure that the transmitted multicast packets are successfully received. Our results indicate that the proposed protocol can indeed lengthen network lifetime, reduce redundant packets, and preserve network reliability.

The fourth work considers a micro-climate monitoring scenario, which usually requires deploying a large number of sensor nodes to capture the environmental information. By exploiting vehicular sensor networks (VSNs), it is possible to equip fewer nodes on cars to achieve fine-grained monitoring. Specifically, when a car is moving, it could conduct measurements at different locations, thus collecting lots of sensing data. To achieve this goal, this work proposes a VSN architecture to collect and measure air quality for micro-climate monitoring in city areas, where nodes' mobility may be uncontrollable (such as taxis). In the proposed VSN architecture, we address two network-related issues: (1) how to adaptively adjust the reporting rates of mobile nodes to satisfy a target monitoring quality with less communication overhead and (2) how to exploit opportunistic communications to reduce message transmissions. We propose algorithms to solve these two issues and verify their performances by simulations. In addition, we also develop a ZigBee-based prototype to monitor the concentration of carbon dioxide gas in city areas.