A ZigBee-Based Localization System for Parking Management
Chen, Kuo Chang
|摘要:||在台灣，至民國99年底自用小客車車輛登記數為564萬輛，如此多的車輛，非常容易造成停車位不敷使用。假日在民眾出遊時，等待車位是必經過程，而停車位一位難求的情況，使得駕駛人需要花費很多時間去找尋或等待車位，遊興跟著大打折扣。本論文實作一組停車場管理系統，透過無線感測網路，協助車主方便的找尋車位，以及讓管理者便於管理掌握車位情況。這個系統採用了TI的OMAP3530，以Embedded Linux作為作業系統。在Qt環境下開發一個應用程式，配合Digi的XBee模組以及Arduino模組，佈建一個以802.15.4的協定的無線傳輸。而中控台伺服器可用於Windows或是Linux上透過XBee傳輸，可將停車位的使用情況在車輛進入時顯示於嵌入式系統上的觸控面板，使駕駛人可以在進入停車場時看到目前車位情況並判斷要將車子停放至哪一可用的停車位，而且中控台伺服器的網頁頁面也可即時更新停車位使用狀態，使管理者便於管控。本論文使用了KNN(k-nearest neighbor)演算法對於車子所停的車位作判斷。以實際的測試，由XBee所接收到由停車位周圍的RSSI(Received signal strength indication)值作演算判斷車子所停的車位，將得到的結果作結果進行準確度的分析。實驗場地為國家衛生研究院竹南分院的地下停車場，實測結果良好，在判斷停車車位成功率大約有九成，可證明此為一可以實際應用的停車解決方案。|
In Taiwan, the registration number of private vehicles is about 5,640,000 by 2010. When drivers are going out in holidays, it is usually hard to find parking spaces. Spending time to find and wait for parking spaces also leads to bad moods. This thesis proposes a parking management system using the wireless sensor network technology. It assists car owners to conveniently find parking spaces and makes the manager easily understand the parking situation. The system was implemented on a platform based on a TI OMAP3530 processor. The sensor node was composed of an XBee module, an Arduino board, and a battery box. All sensor nodes communicate with the platform through ZigBee Protocol. When drivers are entering a parking garage, a center server provides a map showing the status of parking spaces to the platform through an XBee module. The driver can quickly know where the available parking space is. The parking information at the website is also immediately updated, and the manager is easy to know the status of parking spaces. To analyze which parking lot is occupied, a localization approach was developed and verified by real tests. According to the received signal strength indication (RSSI) values from the sensor nodes around the parking space, the embedded system will calculate which parking lot is occupied by k-nearest neighbor (KNN) algorithm. The system was set up in the parking garage of national health research institute (NHRI). The average detection ratio was above 90%, and it can be a practical application as a solution for parking management.
|Appears in Collections:||Thesis|