TSG-RAN WG1 Meeting#44bis
`Athens, Greece, March 27-31, 2006
`
`Source:
`Title:
`Agenda Item:
`Document for:
`
`
`Panasonic
`Random access burst evaluation in E-UTRA uplink
`10.2.3
`Discussion
`
`R1-060792
`
`Introduction
`1.
`Random access burst is used for the initial physical connection on initial cell access, handover and the resource
`allocation when the UE uplink has not been time synchronized. Several discussions on random access burst to
`achieve short initial physical connection setup have also been reported in [4] - [7] . Random access burst sub-
`frame may be composed of a preamble part and a message part. We evaluate the preamble performance. Based
`on the evaluation results, we discuss the inclusion of message part on random access burst.
`2. Discussion
`2.1. Random access burst requirements
`In random access burst structure design, the following requirements have been considered [1] [3] - [10] .
` Reliable acquisition of preamble
` Estimation of arrival timing
` Reduction in the whole process delay
` To minimize the usage of time-frequency resources regarding spectrum efficiency
`The most important requirement of the above is reliable acquisition and estimation of arrival timing because the
`success rate of random access burst attempt should be high enough. The inclusion of message part on random
`access burst has been considered to shorten physical connection setup delay [4] - [7] .
`
`
`2.2. Discussion on preamble length
`In TR [2] , E-UTRA is required to support at least 30km cell size. Therefore, we showed the link budget and
`achievable number of bits per TTI (0.5ms) to estimate how many bits can be contained on random access burst
`in [10] . The result would be useful in the case coverage is critical although the result is still preliminarily. On
`the other hand, we also need the discussion in the case that interference is critical. Ref. [6] reports that
`approximately -13 dB and -18 dB of the average received Es/No were derived from the system level evaluation.
`As mentioned above, the most important random access burst functions are reliable acquisition and estimation of
`arrival timing. For these reasons, first, we evaluate the required preamble length that corresponds to the required
`average received Es/No. Next we discuss the possibility of the inclusion of message part.
`In the preamble evaluation, we assume the followings:
` Random access burst TTI is a multiple of 0.5msec. Preamble, guard time and possibly message part
`share a random access burst TTI
` Random access burst is time/frequency multiplexed with other channels [3] [4] .
`
`-1/7-
`
`
`
`SAMSUNG 1002-0001
`
`

`
`Preamble structure
`A preamble sequence should have a good auto-correlation and good-cross correlation. General chirp-like (GCL)
`sequence has been considered to satisfy these requirements [5] [8] [9] . In our preamble performance evaluation,
`Zadoff-Chu CAZAC sequence [13] , a special case of GCL sequence, is used. RACH preamble structure is
`shown in Figure 1.
`We evaluated 1.25MHz and 5MHz as transmission bandwidth of the random access burst. The preamble
`structure consists of M-times repetition of N=73 (1.25MHz) or N=293(5MHz) CAZAC sequence. Cyclic prefix
`and guard time are also included within a random access burst TTI.
`
`M Repetition (M=3(200u)/ 7(467us) /14 (933us) / 28(1867us))
`
`CP
`
`CAZAC sequence
`N=73(1.25MHz), N=293(5MHz)
`
`CAZAC sequence
`
`CAZAC sequence
`
`Guard
`time
`
`128 sample
`
`512 samples (66.67us)
`Random access burst TTI length = 0.5 / 1.0 / 2.0 ms
`Figure 1 – preamble structure
`
`remaining sample
`
`
`
`
`Performance of preamble
`The simulation parameters are shown in Table 1. As preamble performance evaluation criteria, we used false
`alarm and miss detection probability to the average received Es/No. The definition is as follows:
` False alarm (Pfa): the probability of a particular code being detected when nothing, or different code was
`transmitted
` Miss detection (Pmd): the probability of a particular code not being detected when the code was
`transmitted
`Although time domain preamble detection would also possible, in our evaluation, the RACH preamble detection
`is performed in frequency domain, which is similar to the detection algorithm described in [8] .
`1. Repeated CAZAC sequences of the received signal are combined in time domain.
`2. The combined CAZAC sequence is processed by FFT.
`3. A transmitted CAZAC code is detected by using coherent detection in frequency domain.
`4. A delay profile response is obtained after IDFT processing.
`
`
`5MHz
`
`Table 1 – Simulation parameters
`1.25MHz
`Transmission Bandwidth
`Localized FDMA
`Transmission scheme
`0.5 ms / 1.0ms / 2.0ms
`RACH TTI length
`CAZAC sequence (Zadoff-Chu CAZAC[13] )
`Signature pattern
`Length of CAZAC sequence (N) 73
`293
`3 (total preamble length: 200usec)
`7 (total preamble length: 467usec)
`14 (total preamble length: 933usec)
`28 (total preamble length: 1867usec)
`
`11
`
` transmit antenna, 2 receive antenna (combined non-coherently)
`Coherent detection in frequency-domain
`Preamble detaction in time-domain (after IDFT)
`AWGN
`Typical Urban model, 120km/h
`
`
`
`Repetition factor (M) of
`CAZAC sequence
`Number of multiplexed users
`Antenna configuration
`
`Detector
`
`Channel model
`
`
`-2/7-
`
`SAMSUNG 1002-0002
`
`

`
`Figure 2 and Figure 3 illustrate the miss detection probability (Pmd) to the average received Es/No of 1.25MHz
`and 5MHz bandwidth to achieve the false alarm Pfa = 10-3 under AWGN channel and TU 120km/h, respectively.
`
`
`False alarm Pfa = 10-3
`
`Bandwidth: 5MHz
`Channel: AWGN
` 3 repetition (200us)
` 7 repetition (467us)
` 14 repetition (933us)
` 28 repetition (1867us)
`
`Bandwidth: 1.25MHz
`Channel: AWGN
` 3 repetition (200us)
` 7 repetition (467us)
` 14 repetition (933us)
` 28 repetition (1867us)
`
`100
`
`10-1
`
`10-2
`
`10-3
`
`Miss detection probability (Pmd)
`
`10-4
`-35
`
`-5
`
`-30
`-25
`-20
`-15
`-10
`Average received Es/No [dB]
`
`Figure 2 Miss detection probability (Pmd) to the average received Es/No (AWGN)
`100
`
`False alarm Pfa = 10-3
`
`Bandwidth: 5MHz
`Channel: TU 120km/h
` 3 repetition (200us)
` 7 repetition (467us)
` 14 repetition (933us)
` 28 repetition (1867us)
`
`Bandwidth: 1.25MHz
`Channel: TU 120km/h
` 3 repetition (200us)
` 7 repetition (467us)
` 14 repetition (933us)
` 28 repetition (1867us)
`
`10-1
`
`10-2
`
`10-3
`
`Miss detection probability (Pmd)
`
`10-4
`
`-25
`
`0
`
`-20
`-15
`-10
`-5
`Average received Es/No [dB]
`
`Figure 3 Miss detection probability (Pmd) to the average received Es/No (TU 120km/h)
`
`
`
`
`Target value of the false alarm is  10-3 and the target value of miss detection is  10-2 and 10-3 in WCDMA [11] .
`We think similar target also would be required in LTE. Therefore, if we use the same target values from the
`above results, we can derive the required preamble length to fulfill the average received Es/No. The required
`preamble length in 1.25MHz bandwidth is illustrated in Figure 4 to the average received Es/No to achieve Pmd 
`10-3 and Pmd  10-2 with false alarm Pfa = 10-3. Figure 5 shows the case of 5MHz bandwidth.
`
`-3/7-
`
`SAMSUNG 1002-0003
`
`

`
`Bandwidth: 1.25MHz
` Pmd=10-3 (AWGN)
` Pmd=10-2 (AWGN)
`
` Pmd=10-3 (TU120km/h)
` Pmd=10-2 (TU120km/h)
`
`2
`
`1
`
`0.5
`
`0.2
`
`Required preamble length [ms]
`
`
`
`
`
`0.1
`-25
`
`2
`
`-20
`-15
`-10
`-5
`Average received Es/No [dB]
`Figure 4 Preamble length to Es/No of false alarm probability = 10-3 (1.25MHz)
`Bandwidth: 5MHz
` Pmd=10-3 (AWGN)
` Pmd=10-2 (AWGN)
`
`0
`
` Pmd=10-3 (TU120km/h)
` Pmd=10-2 (TU120km/h)
`
`1
`
`0.5
`
`0.2
`
`Required preamble length [ms]
`
`0.1
`-30
`
`-25
`-20
`-15
`-10
`Average received Es/No [dB]
`Figure 5 Preamble length to Es/No of false alarm probability = 10-3 (5MHz)
`
`-5
`
`
`According to [6] , approximately -13 dB and -18 dB of the average received Es/No were derived from the system
`level evaluation for the ISD 500m and 1732m, respectively, when using open-loop TPC and 5MHz transmission
`bandwidth. Table 2 shows preamble length required for -13dB and -18dB of Es/No under AWGN and
`TU120km/h.
`Table 2 Required preamble length to the average received Es/No (5MHz bandwidth)
`Average received Es/No
`AWGN
`TU-120 km/h
`Pmd = 10-2
`Pmd = 10-3
`5-repetition
`7-repetition
`(333 usec)
`(467 usec)
`14-repetition
`28-repetition
`(933 usec)
`(1867 usec)
`
`-13 dB (ISD=500m)
`
`-18 dB (ISD=1732m)
`
`
`
`Pmd = 10-2
`1-repetition
`(67 usec)
`3-repetition
`(200 usec)
`
`Pmd = 10-3
`2-repetition
`(133 usec)
`4-repetition
`(267 usec)
`
`-4/7-
`
`SAMSUNG 1002-0004
`
`

`
`In this evaluation, only one preamble is transmitted. If multiple preambles are transmitted and multiple
`preambles are also received at the same time, additional preamble length would be required due to multiple
`access interference (MAI).
`
`
`2.3. Random access procedure
`For non-synchronized random access procedure, we introduced the four methods in the Denver meeting [10] .
`We extended the discussion to following five methods. In the figure "preamble" could be randomly chosen
`signature sequence
`
`UEUE
`
`
`
`Node BNode B
`
`
`Random access burstRandom access burst
`
`(preamble + resource request + Data)(preamble + resource request + Data)
`
`Method A
`
`
`
`UEUE
`
`
`Random access burstRandom access burst
`
`(preamble + resource request)(preamble + resource request)
`
`UL resource allocationUL resource allocation
`
`SDCH (UL data)SDCH (UL data)
`
`
`
`Node BNode B
`
`
`
`UEUE
`
`
`Random access burstRandom access burst
`
`(preamble)(preamble)
`
`UL resource allocationUL resource allocation
`
`SDCH (full resource request + UL data )SDCH (full resource request + UL data )
`
`
`
`Node BNode B
`
`Method B
`
`Method C
`
`
`
`Node BNode B
`
`
`
`UEUE
`
`
`Random access burstRandom access burst
`
`(preamble + resource request)(preamble + resource request)
`
`UL resource allocation for resource requestUL resource allocation for resource request
`
`SDCH (full resource request)SDCH (full resource request)
`
`UL resource allocationUL resource allocation
`
`SDCH(UL data)SDCH(UL data)
`
`
`
`Node BNode B
`
`
`
`UEUE
`
`
`Random access burstRandom access burst
`
`(preamble)(preamble)
`
`UL resource allocation for resource requestUL resource allocation for resource request
`
`SDCH (full resource request)SDCH (full resource request)
`
`UL resource allocationUL resource allocation
`
`SDCH(UL data)SDCH(UL data)
`
`Method D
`
`Method E
`Figure 6 Initial resource allocation sequence
`
`
`
` -
`
` Method A
`The random access burst contains preamble, resource request and data. The delay for data transmission
`could be shortest.
`- Method B
`The random access burst contains preamble and resource request. The resource request could tell the amount
`of UE buffer and/or transmitter status. We assume only one or a few bits for this. The allocated amount of
`UL resource could be based on this resource request. The actual data is transmitted after one round trip time
`(RTT).
`- Method C
`The random access burst contains preamble only. The allocated amount of UL resource could be based
`without UE buffer and/or transmitter status. Therefore, the uplink resource allocation is not so accurate and
`could be waste of time-frequency resource in the uplink. The actual data is transmitted after one RTT.
`- Method D
`
`-5/7-
`
`SAMSUNG 1002-0005
`
`

`
`The random access burst contain preamble and resource request. The allocated amount of UL resource in the
`first SDCH would be relatively small because only a few information bits are obtained at Node B. The next
`SDCH contains UL data. The actual data is transmitted after two RTT.
`- Method E
`The random access burst contain preamble and resource request. The allocated amount of UL resource in the
`first SDCH is small because only resource request is transmitted. The next SDCH contains UL data. The
`actual data is transmitted after two RTT. Although the delay of SDCH transmission, the benefit of this
`scheme would be a appropriate amount of time-frequency resource to the second SDCH is based on more
`detailed information of resource request in the first SDCH. Therefore, accurate resource allocation is
`possible.
`
`
`Method A requires different design of random access burst from the others. From the discussion of the previous
`sections, it would be difficult to include a large number of control information bits in a random access burst.
`Method B, C, D and E are almost similar on the design of random access burst.
`
`We prefer method B/D than method C/E, because to include a few bits of the control information is beneficial in
`order to shorten the delay. The difference of B/D and C/E is whether to have a few bits on UE resource status
`(buffer and/or transmission status). On the actual resource status signaling method, there are two approaches.
`One is short message part is included as shown in Figure 7(a). The other is preamble pattern itself is chosen from
`the large number of preamble set shown in Figure 7(b). The choice of preamble pattern itself indicates the
`signaling.
`
`Random access burst
`Preamble part
`Message part
`(signature)
`(resource request)
`(a) two parts structure
`
`Preamble part
`(signature including resource request)
`(b) one part structure
`
`Figure 7 Inclusion of short message in random access burst
`
`
`We don't see so much difference between B and D. The difference between B and D is the amount of uplink
`resource allocation in the first SDCH. Depending on cell level traffic situation, the scheduler could control the
`amount of allocation in the first SDCH. In method B, not only cell level traffic situation, but also the scheduler
`can control the amount of uplink resource based on the resource request from each UE. If UL resource allocation
`is relatively large, method B is applied. If UL resource allocation is relatively small, method D is applied. We
`think this handling of procedure looks useful approach because this enables trade off between delay and
`efficiency.
`As a conclusion, we propose to take method B/D. The difference of B and D can be considered as the difference
`of the scheduler operation.
`
`3. Conclusion
`In this contribution, we evaluated the preamble performance. Based on the evaluation results, we discussed the
`inclusion of message part on RACH. Our current view is that it would be difficult to include a large massage part
`in RACH due to the limitations imposed by the link budget and preamble performance. However, we think that
`the inclusion of a few number of control information bits on random access burst is still beneficial from process
`delay point of view. A small size of message part may be included depending on target Es/No.
`
`
`-6/7-
`
`SAMSUNG 1002-0006
`
`

`
`References
`[1] TR 25.814 V1.0.2, “Physical layer aspects for evolved UTRA”
`[2] TR25.913 V2.0.0, “Requirements for Evolved UTRA and UTRAN”
`[3] R1-051445, Ericsson, “E-UTRA Random Access”
`[4] R1-051391, NTT DoCoMo, “Random Access Transmission for Scalable Multiple Bandwidths in Evolved
`UTRA Uplink”
`[5] R1-060025, Motorola, “RACH Design for EUTRA”
`[6] R1-060047, NTT DoCoMo, NEC, Sharp, “Random Access Transmission in E-UTRA Uplink”
`[7] R1-060181, Qualcomm, “Characteristics of UL Access Channel”
`[8] R1-060152, Nortel, “Consideration on UL RACH scheme for LTE”
`[9] R1-060226, Huawei, “EUTRA RACH preambles”
`[10] R1-060699, Panasonic, “Inclusion of additional data on RACH”
`[11] R1-060061, “LTE L1 related questions to RAN1”
`[12] TR25.104 V6.11.0, “Base Station (BS) radio transmission and reception (FDD) (Release 6)”
`[13] D. C. Chu, “Ployphase codes with good periodic correlation properties,” IEEE Trans. Information Theory,
`vol.18, pp531-532, July 1972.
`
`
`
`-7/7-
`
`SAMSUNG 1002-0007