適用於射頻識別系統之非均等遺失保護群碼技術探討

Abstract

現今的射頻識別系統，是訴求在沒有網路的情況下，也可以藉由讀取每個標籤內記載的資訊去判定是否有標籤遺失，群碼技術與射頻辨識系統的結合就是以此為目標在運行。其運行方式是設計一個二進制的生成矩陣，將標籤做分群，並讓每個標籤記錄分群的相關資訊，使接收端在收到這些標籤後可以依據上面記載的內容去做分析，判定遺失的標籤數目，更甚者可以判定遺失了哪些標籤。而傳統群碼技術對每個標籤的遺失保護能力是相同的，但在現實的運用中，一批貨物裡面每個貨物遺失造成的經濟損失不同，也就是每個貨物之間有重要程度的差異，因此我們希望設計出的群碼技術能夠給予不同標籤不同的錯誤保護能力，此為非均等遺失保護群碼技術。 在本論文中，我們發現部分尾咬迴旋碼的奇偶校驗矩陣，其矩陣的結構為有些列向量因為部分尾咬的作用擁有首尾循環的特性。當我們在對部分尾咬迴旋碼進行擇多邏輯解碼時，擁有首尾循環特性的列向量正交距離比較大，錯誤更正的能力比較好，造成所有列向量的錯誤保護能力不會完全相同。因此本文用一個對稱自正交部分尾咬迴旋碼的奇偶校驗矩陣，做為非均等遺失保護群碼技術的生成矩陣。接著我們分析該生成矩陣中，每個標籤遺失保護的程度與矩陣結構的關係。並模擬用部分尾咬迴旋碼的奇偶校驗矩陣，進行非均等遺失保護擴充群碼技術的編碼與解碼。最後將分析結果與模擬結果做比較，找出在遺失的標籤數目不同的情況下，哪些因素對降低標籤錯誤保護能力的影響較顯著。

Grouping of radio-frequency identification (RFID) tags allows verifying the integrity of groups of objects without external systems. The fundamental idea of grouping is to use binary generation matrices to divide a target collection of RFID tags into overlapping groups, and record the group-related information in each tag. By the technique of grouping, we can learn the number of missing tags and can even identify the missing objects based on the received tags. The original grouping and its extended counterpart, aim to provide equal missing protection for all of the tags. However, there are cases where not all of the tags require the same missing protection. Some goods are regarded as more important than others due to their high commercial value. The concept of unequal-missing-protection (UMP) for grouping and its extended counterpart is that important objects should have higher missing protection levels than other objects. In this thesis, we find that some columns of the parity-check matrices of partial tail-biting convolutional codes have the characteristic of the end-around-shift. When we do the majority-logic decoding of the partial tail-biting convolutional codes, columns with the end-around-shift characteristic have relatively large orthogonal distance, and their ability of error correction are better; so the error protection ability of all columns are not exactly the same. Therefore, we use a parity-check matrix of the systematic, self-orthogonal, partial tail-biting convolutional code as the generation matrix of the unequal-missing-protection extended grouping. Then we analyze the relationship between the missing protection levels of tags and the structure of the generation matrices. We simulate the encoding and decoding of the unequal-missing-protection extended grouping by using the parity-check matrices of partial tail-biting convolutional codes. Finally, we compare the simulation results with the theoretical analysis to find out which factors have a significant effect on the missing protection levels of tags.