利用無線區域網路接收信號強度指示解鎖智慧型手機

Abstract

智慧型手機至今已相當普及,主要是因為計算能力的增強、作業系統穩定 性的提升及應用程式開發工具的開放,各路好手皆可參與手機應用的開發,提供 多樣且吸引人的應用程式,而逐漸使手機成為人們生活中不可或缺的一部份。在 人們對手機已極度依賴的今天,我們每天都得拿起手機查看數十次,無論是檢查 來電、訊息或是電子郵件,一開始都得通過一道解鎖的手續。目前最為普及的解 鎖方式為密碼鎖定,即是在觸控面板上按壓相對應的英文、數字、符號或位置區 塊,但缺點是容易被側錄以及沒效率的重複性動作。後來手機解鎖方式又發展出 聲音、人臉及指紋辨識,這是利用人類的生物特徵來解鎖,但缺點是聲音、人臉 及指紋特徵非常容易被竊取及複製,被竊取後也無法重新設定新的生物特徵,風 險相當高。 本論文研究使用完全不同於以往的方式進行手機解鎖,以 IEEE 802.11 無 線區域網路(WLAN)為基礎,利用手機接收周圍無線存取點(Wireless Access Point, WAP)的廣播訊號,量測 RSSI 的變化量來進行迴歸分析以判斷握持手機 者是否為手機主人。實驗利用市售 Android 智慧型手機分別在公司辦公室及學校 實驗室兩地進行盲測驗證,辨識準確率最佳可達到 98.6%,在未來極具應用價值。

Everywhere smart phones are necessary for everyone today, because computing power improvement, the stable operation system and open application development tools attract everyone to develop multiplicity and attraction of applications. Today everyone depends on smart phones, and they need to check it out many times a day. Whether we check phone calls, messages or e-mails, we need to go through the procedure of unlocking on the smart phone. Currently the most common way to unlock is to enter the password, that is entering the corresponding letters, symbols, numbers or position blocks on the touch screen, but the disadvantage is that the password may be peeped. Another disadvantage is that we need to enter the password repeatedly. Afterward, developments of a voice, personal face and fingerprint recognition, which is the use of biometrics to unlock the screen. We can use biometrics such as personal faces and fingerprints to unlock the screen, but the drawback is that the human face and fingerprint feature is very easy to steal and copy. Our biometrics can not be reset, so in fact, the risk is quite high. In this research, totally different from the previous way of unlocking the phone. We use that smart phones receive broadcast signals around access points over Wi-Fi based on IEEE 802.11 wireless local area network (WLAN), and measuring RSSIs to regression analysis to determine whether the person is an owner or not. The experiment was used off-the-shelf Android phone at my office and my lab. We design to use blind test to pick up data. Identification achieves an accuracy of 98.6%. It will be great value in the future.