SequenceDesignforDLRS.docVIP

Page PAGE4Draft prETS 300 ???: Month YYYY -  PAGE 2/ NUMPAGES 2 - 3GPP TSG-RAN WG1 #49 R1-072028 May 7th-11th, 2007 Kobe, Japan Agenda item: 7.11.1 Source: QUALCOMM Europe Title: Sequence Design for DL RS Document for: Discussion and Decision 1 Introduction DL RS is a product of one orthogonal sequence (1 out of 3) and one two-dimensional pseudo-random sequence (1 out 170)  REF _Ref162167904 \n \h [1]. The three orthogonal sequences have been specified. In this contribution, we make recommendations for the pseudo-random sequence. 2 Analysis Reference sequence within the center 1.25 MHz Before acquiring P-BCH, UE does not know the system bandwidth. But UE may need to use reference signal for Hopping on/off detection and/or antenna information detection Phase reference for P-BCH decoding Thus, the reference pilot sequence within the center 1.25 MHz or 12 tones should be unique regardless of system bandwidth. Cross-correlation properties between different sequences It appears that minimizing the cross-correlation between different pilot sequences is desirable for good channel estimation. But there are other factors making this requirement less important: Due to frequency selectivity, minimizing correlation over a large bandwidth may not be very useful To support different system bandwidths, minimizing correlation over a small bandwidth, e.g., 1.25MHz may be useful In asynchronous systems, pilot tones may collide with data tones. Thus minimizing correlation between different pilot sequences does not help much In synchronous systems, reference signal hopping makes the collision between different pilot tones from different Node Bs less often and/or the collision may come mostly from second tier Node Bs, i.e., the interfering pilot tones may be very weak. In this case, minimizing correlation between different pilot sequences does not help much Thus, minimizing correlation between different pilot sequences is not critical. In the following, we