TSG-RAN WG1 Meeting#45
`Shanghai, China, May 8 - 12, 2006
`
`Source:
`Title:
`Agenda Item:
`Document for:
`
`
`Panasonic
`Random access design for E-UTRA uplink
`11.1.2
`Discussion and Decision
`
`R1-061114
`
`Parameter
`Transmission Bandwidth
`Preamble length
`Guard time
`Signature Pattern
`
`Length of CAZAC sequence (N)
`
`Introduction
`1.
`In this document, we discuss the random access structure as follows. This document only discusses non-
`synchronized random access structure.
` The preamble sequence
` The minimum preamble length
` The minimum bandwidth
` The sequence composition in preamble
` The control information over the preamble part
` The necessity of message part
`2. Random access structure design
`2.1. Preamble sequence
`Random access is a contention based transmission. Therefore, multiple random access bursts from multiple UEs
`could be transmitted simultaneously. It is also good, if multiple random accesses are detected simultaneously at
`E-NodeB. To reduce the collisions among the random access, a common approach is UE randomly chooses one
`out of plural different preambles/signatures. To distinguish random accesses from different UEs at NodeB, a
`sequence with good auto-correlation and good cross-correlation property is required. For these reasons, we
`compare the miss detection probability vs. the average Ep/No among the different type of sequences (i.e. W-
`CDMA preamble sequences, different CAZAC sequences and cyclic-shifted CAZAC sequences).
`Performance of different preamble sequences
`The simulation parameters are shown in Table 1. Preamble performance evaluation criteria used are false alarm
`and miss detection probability to the average Ep/No. The definition is as follows:
` False alarm (Pfa): the probability of a particular code being detected when nothing, or different code is
`transmitted
` Miss detection (Pmd): the probability of a particular code not being detected when the code is transmitted
`Table 1 Simulation parameters
`Value
`1.25MHz (Allocated bandwidth: 1.024MHz)
`Approximately 400 usec
`Approximately 100 usec
`- W-CDMA (truncated)
`- CAZAC sequence (Zadoff-Chu CAZAC[20] )
`- W-CDMA (400 symbols: 16 signature * 25 repetition)
`- CAZAC (401 symbols)
`- Cyclic-shifted CAZAC (401symbols, shift duration: 50usec)
`1, 2, 4, 8, 12, 16
`1 Tx antenna, 2 Rx antennas (power profiles are combined)
`Matched filtering in time domain. See Appendix.
`16
`6-path Typical Urban 120km/h
`
`Number of multiplexed preambles
`Antenna configuration
`Detector
`Number of detector
`Channel model
`
`
`Figure 1 shows the miss detection probability (Pmd) against the average Ep/No of each preamble sequence to
`achieve the false alarm Pfa = 10-3 under TU 120km/h. The miss detection probability against the Ep/No is always
`satisfied in Pfa = 10-3. The result reflects that the false alarm probability is fluctuated due to mutual interference
`between preambles when plural preambles are transmitted.
`
`-1/5-
`
`SAMSUNG 1003-0001
`
`

`
`From the evaluation, both CAZAC sequence and cyclic-shifted CAZAC sequence show better detection
`performance compared with the truncated WCDMA preamble sequence. Eight cyclic-shifted CAZAC sequences
`mixed have similar performance with only one CAZAC sequence. Moreover, the performance in 8 cyclic-shifted
`a CAZAC sequences and 4 cyclic-shifted other CAZAC sequences mixed have similar to 4 different CAZAC
`sequences mixed. Therefore, cyclic-shifted CAZAC sequence has superior performance among compared
`sequences. This aspect is also discussed in [14] .
`As the results, we propose to choose cyclic-shifted Zadoff-Chu CAZAC as preamble sequence mainly. In
`addition, to have more signatures, we also propose to use different Zadoff-Chu CAZAC sequence.
`100
`
`False alarm (Pfa) = 10-3
`
`WCDMA (1signature)
`WCDMA (2signatures)
`WCDMA (4signatures)
`
`CAZAC (1 signature)
` (k=1)
`CAZAC (2 signatures)
` (k=1,2)
`CAZAC (4 signatures)
` (k=1,2,3,4)
`CS-CAZAC (8 signatures)
` (k=1, m=1to8)
`CS-CAZAC + CAZAC (12 signatures)
` (k=1, m=1to8) + (k=2, m=1to4)
`CS-CAZAC + CAZAC (16 signatures)
` (k=1, m=1to8) + (k=2, m=1to8)
`
`10-1
`
`10-2
`
`10-3
`
`Miss ditection probability
`
`10-4
`10
`
`
`
`6-pathTU 120km/h
`15
`20
`Average Ep/No [dB]
`Figure 1 Miss detection probability (Pmd) to the average Ep/No (TU 120km/h)
`
`NOTE:
`CS-CAZAC: cyclic-shifted CAZAC
`- k is the index of CAZAC sequence
`- m is the index of cyclic shift
`
`25
`
`
`
`2.2. Preamble length
`Approximately 300 usec preamble length is required for ISD=500m and approximately 900 usec is required for
`ISD=1732m to achieve Pmd = 10-3 on CDF = 5% under TU 120km/h from the preamble detection performance
`in [13] . In the document, power control scheme assumed is relatively simple one. If more sophisticated one is
`assumed, the averaged received SINR at CDF = 5% would be further improved. In addition, more sophisticated
`preamble detectors in [15] [16] improves the preamble detection performance. These two aspects would allow
`reducing the required preamble length. Therefore, we propose to have two preamble lengths, around 400 usec
`and around 800 usec.
`
`2.3. Minimum bandwidth
`We propose the minimum bandwidth (BW) of random access burst is 1.25MHz. More than 1MHz BW would be
`required in order to obtain 1 usec time resolution for the uplink time alignment [19] . If only rough resolution is
`obtained in random access procedure, timing alignment control after random access procedure would get
`complicated.
`In addition, sufficient number of symbols of the CAZAC sequence is required to eliminate mutual interference
`among preamble signatures. Therefore, we propose 1.25MHz as the minimum bandwidth.
`
`2.4. Sequence composition in preamble
`In the previous sections, we discussed the preamble sequence, the preamble length and the minimum bandwidth.
`Next topic is how to fulfill the possible preamble field using preamble sequence. Two approaches have been
`proposed. One is composed of multiple short CAZAC sequences [15] [16] . The other is one long CAZAC
`sequence [19] . For the decision among two, following aspects should be considered.
`- Mutual interference among preambles
`- Reuse factor of CAZAC sequence
`- The possibility to transmit control information
`- Decoder complexity
`
`-2/5-
`
`SAMSUNG 1003-0002
`
`

`
`
`Mutual interference among preambles
`Multiple short CAZAC sequence approach suffers more mutual interference among preambles. In addition, as
`we saw the evaluation in section 2.1, cyclic-shifted CAZAC sequence has superior performance. But cyclic-
`shifted CAZAC sequence requires relatively long sequence. Therefore, long CAZAC sequence is better than
`multiple short CAZAC sequence on this aspect.
`Reuse factor of CAZAC sequence
`The longer CAZAC sequence has a benefit to have bigger reuse factor of sequence management with less inter-
`cell interference when cell planning aspect is considered [19] . Therefore, long CAZAC sequence is better than
`multiple short CAZAC sequence on this aspect.
`The possibility to transmit control information
`To have a few number of control information bits on random access burst allows of an more efficient uplink and
`downlink resource utilization after random access attempt. In the case control information is mapped on the
`preamble part, control information including random ID is mapped to different signatures one by one. This
`means the more control bits are contained, if the larger number of signatures is used in one cell. Therefore, the
`required length of CAZAC sequence increases when more number of control bits is used. In addition, the length
`of CAZAC sequence further increases when bigger reuse factor are used. The number of different CAZAC
`sequences used by one cell is shown in Table 2. The number in ( ) shows the case four cyclic-shifted sequence
`are generated for each CAZAC sequence.
`Table 2 the number of CAZAC sequences used in one cell
`Number of control information
`3 cell reuse
`4 cell reuse
`bits (including random ID)
`5 bits
`6 bits
`7 bits
`8 bits
`9 bits
`
`7 cell reuse
`
`224 (56)
`448 (112)
`896 (224)
`1792 (448)
`3584 (896)
`
`96 (24)
`192 (48)
`384 (96)
`768 (192)
`1536 (284)
`
`128 (32)
`256 (64)
`512 (128)
`1024 (256)
`2048 (512)
`
`
`Discussion
`From above discussion, long CAZAC sequence is preferred option. From the previous sections, we proposed 400
`usec as the minimum preamble length and 1.125MHz (90% of 1.25MHz) as the minimum preamble bandwidth.
`Therefore, the maximum number of symbols contained in the preamble part is around 450 symbols.
`
`
`Minimum random access TTI (=0.5msec)
`
`400us
`Preamble part (N=449 CAZAC sequence)
`(6-8 control information bits incl. random ID)
`
`100us
`
`Guard time
`
`
`
`Figure 2 proposed the non-synchronized random access structure
`
`
`We propose the N=449 (prime number) cyclic-shifted CAZAC sequences with also use different CAZAC
`sequences for the preambles. For supporting larger cell size, repeating this sequence twice (i.e. 800 usec) can be
`used.
`According to this design, up to 8 control information bits including random ID can be transmitted on the
`preamble part with 7 cell reuse. A fewer usage of code sequence alleviate the decoder complexity. With also
`taking into account complexity aspect, we propose the number of control information bits contained in the
`preamble is around 6 bits.
`
`2.5. Control information over the preamble part
`We propose the followings control information is transmitted in non-synchronized random access preamble part.
`
`-3/5-
`
`SAMSUNG 1003-0003
`
`

`
` Random ID: To avoid collisions and to distinguish random access attempt from different UEs.
` Access type and buffer status: To allocate appropriate first uplink resource corresponding to the access
`reasons. One example is to distinguish among initial access/TA-update, handover, recovery of the
`synchronization in LTE_ACTIVE with bigger buffer size and recovery of the synchronization in
`LTE_ACTIVE with smaller buffer size
` UE Tx power head room or Downlink CQI: To perform link adaptation and/or power control for
`allocated uplink/downlink resource.
`Example of possible mapping usage of 6 bits is shown in table 3. Similar way of mapping is also proposed in
`[15] .
`
`Table 3 Example of propose control information mapping to signatures
`Tx power head room
`Cause/Access type
`Signature ID (=Random ID)
`(case of 64 signatures)
`
`Large Tx power head
`room
`
`Middle Tx power head
`room
`
`Small Tx power head
`room
`
`No Tx power head room
`
`
`
`Initial access/TA-update
`Handover
`LTE_ACTIVE(small buffer size)
`LTE_ACTIVE(large buffer size)
`Initial access/TA-update
`Handover
`LTE_ACTIVE(small buffer size)
`LTE_ACTIVE(large buffer size)
`Initial access/TA-update
`Handover
`LTE_ACTIVE(small buffer size)
`LTE_ACTIVE(large buffer size)
`Initial access/TA-update
`Handover
`LTE_ACTIVE(small buffer size)
`LTE_ACTIVE(large buffer size)
`
`1-3
`no allocation
`4-6
`7-9
`10-13
`no allocation
`14-17
`18-21
`22-26
`no allocation
`27-31
`32-36
`37-45
`46-54
`44-64
`no allocation
`
`2.6. Necessity of message part
`If more than 6-8 control bits are required to be transmitted on random access burst, the message part has to be
`associated with the preamble part. However, in that case, the preamble part and message part should support the
`following properties.
` Channel estimation for coherent detection by the preamble part
` Message part should have similar BLER with miss detection probability of the preamble part.
` Message part should have similar collision avoidance performance with that of preamble part.
`In order to achieve the above requirements, the longer associated message part might be required [17] . This
`consumes more uplink radio resources. Therefore, the trade-off between the merit of associating message part
`and the demerit of radio resource expense should be carefully considered.
`3. Conclusion
`We propose the following random access burst.
` Zadoff-Chu CAZAC sequence for the preamble sequence
` Both of cyclic-shifted CAZAC and different CAZAC sequence is used.
` Preamble lengths is around 400 usec and around 800 usec
`
`1.25MHz is the minimum bandwidth
` One large CAZAC sequence for example N=449 is used to compose preamble sequence.
` The following control information is mapped on the CAZAC preamble signatures.
` UE Tx power head room or downlink CQI
` Access type and buffer status
` Random ID
`
`-4/5-
`
`SAMSUNG 1003-0004
`
`

`
`
`References
`[1] TR25.814 V1.2.2, “Physical layer aspects for evolved UTRA”
`[2] TR25.913 V2.0.0, “Requirements for Evolved UTRA and UTRAN”
`[3] TR25.104 V6.11.0, “Base Station (BS) radio transmission and reception (FDD) (Release 6)”
`[4] R1-051058, Texas Instruments, “RACH Preamble Design”
`[5] R1-060047, NTT DoCoMo, NEC, Sharp, “Random Access Transmission in E-UTRA Uplink”
`[6] R1-060152, Nortel, “Consideration on UL RACH scheme for LTE”
`[7] R1-060161, Panasonic, “Inclusion of additional data on RACH”
`[8] R1-060181, Qualcomm, “Characteristics of UL Access Channel”
`[9] R1-060226, Huawei, “EUTRA RACH preambles”
`[10] R1-060376, Texas Instruments, “RACH preamble design for E-UTRA”
`[11] R1-060387, Motorola, “RACH Design for EUTRA”
`[12] R1-060541, Huawei, “Some Considerations for Random Access Frame Design”
`[13] R1-060792, Panasonic, “Random access burst evaluation in E-UTRA uplink”
`[14] R1-060797, Huawei, “RACH design for E-UTRA”
`[15] R1-060786, NTT DoCoMo, “Random Access Channel Structure for E-UTRA uplink”
`[16] R1-060908, Nortel Networks, “On the performance of LTE RACH”
`[17] R1-060909, Nortel Networks, “Consideration on the issues of LTE RACH”
`[18] R1-060992, NTT DoCoMo, “Investigations on Random Access Channel Structure for E-UTRA Uplink”
`[19] R1-060998, Ericsson, “E-UTRA Random Access Preamble Design”
`[20] D. C. Chu, “Ployphase codes with good periodic correlation properties,” IEEE Trans. Information Theory,
`vol.18, pp531-532, July 1972.
`
`
`
`Appendix: Preamble detection algorithm
`Two receiver antenna diversity reception is used. The 16 different power delay profiles are measured by the 16
`matched filters corresponding to preamble sequences in each branch and then combined. Figure A illustrates the
`preamble detection method. The window size of the peak detection of the delay profile is set to 100usec for
`WCDMA preamble and CAZAC preamble. The window size for Cyclic-shifted CAZAC preamble is 50 usec to
`evaluate the detection performance up to 8 cyclic-shifted CAZAC sequences. Noise level is measured from the
`delay profile but the samples larger than Threshold A are not used for noise level calculation. Threshold B is the
`preamble detection threshold from the calculated noise level plus an offset value. The offset value is adjusted to
`achieve 0.1% false alarm probability. The maximum peak power is compared to Threshold B.
`
`
`Window size for peak detection
`- 100usec for WCDMA preamble and CAZAC preamble
`- 50usec for cyclic-shifted CAZAC preamble
`
`Peak power
`
`Threshold B: preamble detection threshold
`
`Threshold A: delayed signal detection threshold
`
`Samples (>threshold A) are not
`used for noise level calculation
`Figure A Output signal of matched filter and preamble detection algorithm
`
`Window size for noise level
`calculation = 400 samples
`
`Calculated noise level
`(including interference)
`
`
`
`
`
`-5/5-
`
`SAMSUNG 1003-0005