正交分頻多工系統使用恆定振幅序列進行隨機存取前置訊號偵測

Abstract

在上行蜂巢環境中，隨機存取前置訊號偵測 (random-access preamble detection; RAPD) 是通過閾值比較去識別可能接收到的多個攜帶著任意選擇的恆定振幅 (constant-amplitude; CA) 前置序列的正交分頻多工 (orthogonal frequency-division multiplexing; OFDM) 信號。在獨立的萊斯或瑞利多路徑通道以及多個接收到的OFDM信號之間存在時間偏移的情況下，本論文針對誤報、錯誤和誤識別機率對閾值法RAPD的性能進行了分析研究。當前置訊號序列集合包含相互正交的CA序列以及盡可能多的最小循環移位距離ϖ_min-允許的循環移位恆定振幅 (cyclically-shiftable constant-amplitude; CSCA) 序列時，RAPD系統表現出較少的序列間干擾影響，這些ϖ_min-允許的CSCA序列可以彼此在時域以循環位移ϖ_min產生。 因此在本論文中，分別展示了由正交的 ϖ_min-允許的CSCA序列子家族組成的修改相位模型分配 (modified phase-model-assigning; MPMA) 序列家族，以及由非正交的 ϖ_min-允許的CSCA序列子家族組成Zadoff-Chu序列家族家族的RAPD系統性能特徵。在限制誤報和錯誤機率上限的條件下，採用MPMA序列家族的系統能提供比採用Zadoff-Chu的系統更小的誤識別機率。

In uplink cellular environments, random access preamble detection (RAPD) is achieved by identifying possibly multiple received requesting orthogonal frequency-division multiplexing (OFDM) signals carrying arbitrarily-chosen constant-amplitude (CA) preamble sequences through threshold comparison. Such threshold-based RAPD is analytically studied in terms of false alarm, outage, and false identification probabilities for the receiver operating under independent Rician/Rayleigh multipath channels and in the presence of time offsets among multiple received requesting OFDM signals. The RAPD system is shown to operate under less influence of inter-sequence interference when the preamble sequence set consists of mutually orthogonal CA sequences and contains as many ϖ_min-permissible cyclically-shiftable constant-amplitude (CSCA) sequences as possible, where ϖ_min-permissible CSCA sequences are cyclically shifted from each other with a minimum cyclic shift distance ϖ_min in time domain. Thus, the modified phase-model-assigning (MPMA) sequence family comprised of orthogonal ϖ_min-permissible CSCA sequence subfamilies, and Zadoff-Chu, which is comprised of nonorthogonal ϖ_min-permissible CSCA sequence subfamilies, are demonstrated for RAPD performance evaluation. While constraining upper false-alarm and outage probability bounds, the system adopting the MPMA sequence family is shown to provide much smaller false-identification probability than the system adopting the Zadoff-Chu sequence family.