`
`Fast Acquisition Scheme and Implementation of PRACH in WCDMASystem
`
`Ling Qiu, Yongjia Huang, Jinkang Zhu
`University of Science and Technology of China, Box 4, Hefei, Anhui, 230027, China
`
`Abstract: The performance and implementation of
`PRACH(physical random access channel) acquisition in
`WCDMA systemis investigated in this paper. The analysis
`shows that
`the conventional methods are not satisfying.
`Thus we proposed the quasi-matched filter acquisition
`scheme of PRACH preamble which based on_
`fast
`Hadamard transform. And we implement this method by
`
`hard ware in the practical WCDMA field try system. The
`simulation and test results show that the proposed scheme
`achieves
`the
`following
`performance:
`the
`detection
`probability with £,/N, =7db does not less than 95%, and
`the mean acquisition time is less than 1.33ms.
`
`I.
`
`INTRODUCTION
`
`Pesudo-noise(PN) code acquisition is the first action
`of any direct-sequence code division multiple access(DS-
`CDMA)
`system. Acquisition
`refers
`to
`the
`coarse
`synchronization of the received PN sequence and the
`locally generated PN sequence within a fraction of the chip
`duration of
`the
`code
`sequence. Fast
`and effective
`acquisition is one of the key techniques
`in CDMA
`communication system. And the acquisition performance
`can even be a limitation of the capacity of a system
`(acquisition-based capacity)[1].
`
`Acquisition can be performed and classified in several
`ways[2-3]. One way to separate acquisition methods is
`based on correlation measurementprinciple, which is used,
`active or passive measurement or a combination of the
`two[4].
`In the passive method, a filter matched to the
`spreading code is used. But
`in the active method,
`the
`received signal
`is multiplied with the locally generated
`replica of the spreading code, and the result is integrated
`over some observation interval. The multiplication and
`integration is performed step-by-step for each code phases
`to be tested. Usually, serial, complete parallel. store parallel
`and hybrid parallel search(S-7] approach can be used for
`the detection of the code phases to be tested. However,
`
`serial search with correlation integration method seldom
`used for fast acquisition.
`
`random access
`the physical
`In WCDMA systems,
`channel (PRACH) has one or more preambles and each
`
`preamble is only 4096 chips. The single acquisition time is
`limited in 5120chips. Thus, there is high requirement for
`fast acquisition of RACH preamble. At the sametime, there
`are a total of 16 signatures of RACH preamble part. This
`
`further increases the difficulty of the acquisition scheme. In
`addition, there need two sets of identical acquisition device
`
`for 1/Q branches. Thus, the acquisition efficiency must be
`compareto the matchfilter.
`
`Based on above discussion and aiming at to achieve
`good performance and reduce the hardware resource at the
`same time, we propose a novel acquisition scheme. The
`simulation and test results show that the proposed scheme
`can be easy implement
`in practical WCDMA field try
`system and achieve high performance.
`
`Il. WCDMA PRACH DESCRIPTION
`
`In WCDMA system, the random-access transmission
`
`on a Slotted ALOHA approach with fast
`is based.
`acquisition indication. There are |5 access slots per two
`physical frames and they are spaced 5120 chips apart. See
`The
`of
`the
`reference[8].
`structure
`random-access
`
`transmission is also shown in reference{8]. The random-
`access transmission consists of one or several preambles of
`length 4096 chips and a message of length 10 ms or 20 ms.
`Each preamble is of length 4096 chips and consists of 256
`repetitions of a signature of length 16 chips. There are
`maximum 16 available signatures. The baseband modulator
`illustrated
`as
`figure
`1.
`for RACH is
`The
`binary
`preamble a(&)is modulated to get
`the complex valued
`preamble 5(k).
`
`b(k) = a(k)e
`
`ie k
`wis
`
`lg =0,1,2,...,.4095
`
`(1)
`
`0-7803-7005-8/01/$10.00 © 2001 IEEE
`
`1701
`
`(cid:50)(cid:74)(cid:86)(cid:13)(cid:1)(cid:45)(cid:15)(cid:13)(cid:1)(cid:41)(cid:86)(cid:66)(cid:79)(cid:72)(cid:13)(cid:1)(cid:58)(cid:15)(cid:1)(cid:66)(cid:79)(cid:69)(cid:1)(cid:59)(cid:73)(cid:86)(cid:13)(cid:1)(cid:43)(cid:15)(cid:13)(cid:1)(cid:19)(cid:17)(cid:17)(cid:18)(cid:13)(cid:1)(cid:48)(cid:68)(cid:85)(cid:80)(cid:67)(cid:70)(cid:83)(cid:15)(cid:1)(cid:39)(cid:66)(cid:84)(cid:85)(cid:1)(cid:66)(cid:68)(cid:82)(cid:86)(cid:74)(cid:84)(cid:74)(cid:85)(cid:74)(cid:80)(cid:79)(cid:1)(cid:84)(cid:68)(cid:73)(cid:70)(cid:78)(cid:70)(cid:1)(cid:66)(cid:79)(cid:69)(cid:1)(cid:74)(cid:78)(cid:81)(cid:77)(cid:70)(cid:78)(cid:70)(cid:79)(cid:85)(cid:66)(cid:85)(cid:74)(cid:80)(cid:79)(cid:1)(cid:80)(cid:71)(cid:1)(cid:49)(cid:51)(cid:34)(cid:36)(cid:41)(cid:1)(cid:74)(cid:79)(cid:1)(cid:56)(cid:36)(cid:37)(cid:46)(cid:34)(cid:1)(cid:84)(cid:90)(cid:84)(cid:85)(cid:70)(cid:78)(cid:15)(cid:1)(cid:42)(cid:38)(cid:38)(cid:38)(cid:1)(cid:22)(cid:21)(cid:85)(cid:73)(cid:1)(cid:55)(cid:70)(cid:73)(cid:74)(cid:68)(cid:86)(cid:77)(cid:66)(cid:83)(cid:1)(cid:53)(cid:70)(cid:68)(cid:73)(cid:79)(cid:80)(cid:77)(cid:80)(cid:72)(cid:90)(cid:1)
`Qiu, L., Huang, Y. and Zhu,J., 2001, October. Fast acquisition scheme and implementation of PRACH in WCDMAsystem. IEEE 54th Vehicular Technology
`Conference. VTC Fall 2001. Proceedings (Cat. No. 01CH37211) (Vol. 3, pp. 1701-1705). IEEE.
`(cid:36)(cid:80)(cid:79)(cid:71)(cid:70)(cid:83)(cid:70)(cid:79)(cid:68)(cid:70)(cid:15)(cid:1)(cid:55)(cid:53)(cid:36)(cid:1)(cid:39)(cid:66)(cid:77)(cid:77)(cid:1)(cid:19)(cid:17)(cid:17)(cid:18)(cid:15)(cid:1)(cid:49)(cid:83)(cid:80)(cid:68)(cid:70)(cid:70)(cid:69)(cid:74)(cid:79)(cid:72)(cid:84)(cid:1)(cid:9)(cid:36)(cid:66)(cid:85)(cid:15)(cid:1)(cid:47)(cid:80)(cid:15)(cid:1)(cid:17)(cid:18)(cid:36)(cid:41)(cid:20)(cid:24)(cid:19)(cid:18)(cid:18)(cid:10)(cid:1)(cid:9)(cid:55)(cid:80)(cid:77)(cid:15)(cid:1)(cid:20)(cid:13)(cid:1)(cid:81)(cid:81)(cid:15)(cid:1)(cid:18)(cid:24)(cid:17)(cid:18)(cid:14)(cid:18)(cid:24)(cid:17)(cid:22)(cid:10)(cid:15)(cid:1)(cid:42)(cid:38)(cid:38)(cid:38)(cid:15)
`
`1
`
`APPLE 1015
`
`1
`
`APPLE 1015
`
`

`

`
`
`x Atk)
`+
`1+
`Scrambling code
`I
`Lypch-sl..
`2
`—__—————'
`4096chips
`
`+ atk) .
`
`gp!
`16chips
`Hm: Theset oflength 16 Walsh Hadamard code
`
`Fig.1 Structure of preamble & scrambling code and baseband modulator for RACH preamble
`
`Ill. Quasi-Matched Filter Scheme Based on Fast
`HadamardTransform
`
`At the bsae station, the code phase of the preamble is
`retarded by a roundtrip delay compared with that of the
`
`pilot signal[8]. Hera we assume that the code phase which
`need to be tested equals 400(200chips). We also assume
`that the correlation period is 1024 chips. In this case, the
`serial search method.is not suitable distinctly. If the hybrid
`
`parallel[7] acquisition approach is adopted here. To limit
`the acquisition time within 5120chips, we appoximately
`assume that the 400 phase to be tested is divided into 14
`
`it should finish the search of
`sub-phase intervals. Thus,
`400/30 =14 unknown phase within each sub-phaseinterval.
`
`the total parallel
`16 signatures and VQ branches,
`For
`correlators needed to detect
`the unknown phase are
`16x 2x30 =960. It is impractical for implementation of such
`huge
`correlator
`arrays.
`If conventional match filters
`method[9] is used for PRACH, there need two sets of 16
`
`match filters for I/Q branches of 16 signatures. As we know,
`a digital match filter actually is a multi-stage parallel
`accumulator. And the number of accumulators relate to
`
`correlation length. At the correlation period is 1024chips,it
`needs 1023 accumulators to implement such a match filter.
`
`is disastrous for system resource to implement
`It
`method (total 16x 2x 1023 = 32700 accumlators).
`
`this
`
`Based on the above discussion and the property of
`PRACH, we propose the
`following new acquisition
`detector structure.
`
`the preamble is
`From figure 1, we can find that
`generated by a signature consists of 256 repetitions of a
`length 16 signature and scrambling code. The signature is
`from the set of 16 Hadamard codes of length 16. For the
`samecell, the scrambling code ofall mobile terminals used
`
`is the same. Thus, fast Hadamard transform (FHT) can be
`used to do the correlation of the preamble. The acquisition
`time when using FHT can be shortened. Furthermore, FHT
`can be implemented by original position operation and this
`cannot expend too much hardware resource. FHT is only
`change
`the modality of
`correlation
`operation,
`the
`conventional structure of the acquisition system can still be
`used.
`
`Here we propose the acquisition system of quasi-
`matched filter scheme based on the FHT.
`It
`is shown as
`
`figure 2. According to the figure 1, a non-coherent QPSK
`detector is used for for each FHT branch. The acquisition
`system consists of
`two sets of L(16) FHT which
`
`corresponding to the 16 signatures for each I/Q branches
`
`and a maximum selection unit and a threshold comparing
`
`part. Ifthe maximum correlation value Z,,.,
`
`max
`
`is larger than
`
`the threshold, then the corresponding phase is considered as
`correct phase and the system turns into tracking state.
`Otherwise,
`it goes back to the state of searching another
`unknown phase. The threshold setting principles we used
`here is Neyman-Pearson criterion. That is at the condition
`of constant
`false alarm rate(CFAR),
`to maximize the
`
`detection probability[5,9].
`
`1702
`
`2
`
`

`

`
`Far | accumlater [>
`square 1
` branch scrambling
`
`going to
`Q
`FHT — accumlaror
`mal
`square
`racking
`Maximum
`is
`seleciion
` Q branch scrambling
`
`code generator
`avcumlator Ly]
`
`code generator
`
`r(t}
`
`
`
`i
`
`search controller
`
`
`
`
`
`
`
`
`
`
`sqjuare 1
`rt square
`if
`
`
`
`accumlator
`
`
`Fig. 2 a non-coherent QPSK acquisition system based on the FHT
`
`In figure 2, r(t) is the received baseband signal in base
`station. & is the threshold set for the search mode and ¢,
`is the threshold set for the verification mode.
`
`into
`In practical system, the scrambling code is put
`memory. The input data stream is first put into the register
`and the size of the register is chosen as two times as the
`correlation length. Thus,
`the ping-pang operation can be
`done. Every time 16chips data is read out and scrambling
`code is de-spreaded. Then the results will be fed into FHT
`and accumulator. Finally the corresponding results of /Q
`branches are squared and summed. The decision variable
`can be obtained.
`
`For one FHT, 4 accumlators are needed. To complte
`
`the detection of 400 phaseto be tested, the hybrid paraHlel[7]
`acquisition approachis also adopted here and the 400 phase
`to be tested is divided into 200 sub-phaseintervals. Thus, it
`should finish the search of 400/200=2 unknown phase
`
`Table | parameters used for simulation
`3.84Meps
`CFAR
`Threshold set method
`__ETSI vehicular A[1i
`Multipath model
`
`within each sub-phase interval. Thus, for this system. there
`Multipath fading|| Rayleigh
`are total
`(16x 4+ 16)x 2x 2 = 320 accumlators. This is much
`
` Chip rate
`
`The simulation resuits are as follows.
`
`guarantee the required performance.
`
`IV.
`
`Performance analysis and simulation results
`
`Based on the above discription, we do the simulation on
`
`mean acquisition time and detection probability. The
`simulation parameters are listed in table!. The correlation
`length is set as 1024 chips. The penalty factor for the false
`alarm is selected as two, i.e., the penalty time is 2T s. The
`threshold parameters and €,
`are adjust
`to meet
`the
`detection probability and minimum meanacquisition time
`of
`the
`requirements. The
`another
`two
`parameters
`verification mode are quoted from [10]. In the conventional
`hybrid parallel method, the parallel branch is set as 100.
`Andin each paralle! branch, there are 4 phaseto betested.
`
`the proposed meng
`
`aeeseye
`
`
`bo
`oo
`
`O8
`
`LO* “fo
`leconvxyk
`hybridmemywbodif
`03eon
`
`2 v
`
` TT 7
`ot
`Tr
`to
`ab 4213
`ho)
`2} pfele pk
`Eb/Noidby
`
`less than that used in MF method and conventional hybird
`
`parallel method. Hadamard sequence used here is only 16
`chips, the original position operation doesn’t need. We can
`usé streamline structure to do this operation and the
`acquisition time can be further shortened. This streamline
`structure just is a match filter for the Hadamard sequence
`
`implementation. To
`In practical
`with length of 16 chips.
`increase the process gain.
`the correlation length should
`greater than I6chips. It can select as N times of 16chips and
`N is an integer. The desired correlation result
`is
`the
`accumulation of N consecutive output value of the match
`filter. Here we choose N as 64 and such selection can
`
`1703
`
`Pa
`
`3
`
`

`

`Pr
`
`4
`
`ole
`0.09
`0.08
`0.07
`0.06
`0.05
`0.04
`0.03
`0,02
`0.01
`
`0,00
`
`
`
`10000,
`9000
`ROOTE
`
`am
`
`&$
`
`7000+
`s
`cova
`2
`5 5000
`B -an00
`
`9
`=
`3
`
`= 3000
`“theee
`
`hybridiparallei metho
`
`‘<bee
`2000 4-
`So
`+ ke
`‘
`1000
`
`0
`
`00
`O81
`862
`03
`U4
`08
`C6
`OF
`O08
`09
`LO
`Pd
`
`
`
`
`
`
`
`
`4000
`
`
` meanacquisitiontime(chips)
`
`
`10000
`9000
`ROO
`7000
`SS
`athe convéntional
`6000
`
`apeeTeasmets
`fom
`5000
`9 Bee .
` 3000
`2000
`1000
`
`Fig.3
`
`performanceofsingle dwell mode
`
`Fig.4 performance of double dwell mode
`
`We can see from fig.1 that the detection probability(Pd)
`will
`increase and the false-alarm probability(Pf) decrease
`with the inceraseing ofsignal to noise ratio.
`
`~
`
`4.0
`Oot
`OR
`OF
`0.6
`0s2
`mua
`93
`02
`01
`ao
`
`
`
`Eb/No{db}
`
`We can see from fig.3 and fig.4 that when the double
`dwell search method is used, a relatively low threshold &,
`can be used to increase the detection probability, and the
`
`verification mode operation can reduce the probability of a
`false-alarm event: therefore, the mean acquisition time can
`be greatly reduced.
`
`V. Hard ware implementation and performanceresults
`
`We use Alter FPGA 20K200EFC-3 to evaluate the
`hardware scale of the PRACH acquisition system. Total
`.
`:
`11740 logic cells and 142336ESB are used. The diagram
`8
`
`for the FPGA implementation is shown in figure 5.
`
`a(t)
`
`ee en vt
`
`\ branch
`data registor
`Corot
`LV
`
`FHT mi ) | [thes
`.
`“IR/Wdata address
`!
`.
`y
`a ~ read scrambling =~ = "~~ Cceanbipal —__.
`
`address generater of=code.addressp.; scramble code ' complex scramble = compare .
`
`
`read/write RAM 4 | Benerater
`code despread
`a
`‘par
`~
`‘decision
`_»
`Sa
`oa Te 0
`-
`Iwdata address
`Crcranhhing.Q
`te car 5 | _f
`output
`: Me
`ee
`Q branch data
`ed
`registor
`
`rte)
`
`Fig.5 the hardware structure of PRACH acquisition scheme
`
`1704
`
`4
`
`

`

`
`
`One of the branches of the process of PRACH
`acquisition
`is
`described
`above.
`It’s
`hardware
`implementation structure mainly contains the following
`modules.
`
`1. DATA_REGISTER
`This module receives signal from A/D, and put the data
`
`into register. The size of the register is two times as
`correlation length. This unit
`is the main limitation for
`FPGAresource.
`
`2. ADDRESS_GENERATOR
`This module generates the R/W(read/write) address for
`both I and Q branches data registers. It also generates the
`read address of scrambling code generator.
`3. SCRAMBING _CODE GENERATER
`This module generates the scramble code that is used for
`current cell according to initial state that CPU informs.
`4. COMPLEX SCRAMBLE _CODE DESPRADING
`The complex scramble code is de-spread in this module.
`The scramble code multiples the input data step-by-step
`and the results are output.
`5. FHT
`
`from
`come
`I/Q branches
`results of
`The multiple
`COMPLEX
`SCRAMBINGCODE
`DESPRADING
`module are separate fed into FHT modules. 16 parallel
`
`Reference
`
`1. Mohamed G. El-Tarhuni, Asrar U. H. Sheikh, An
`
`Adaptive Filtering PN Code Acquisition Scheme with
`Improved Acquisition Based Capacity in DS/CDMA.
`proc. of PIMRC’ 98, pp 1486-1490, 1998
` Polydoros, On the Synchronization Aspects of Direct-
`Ph.D
`Sequence
`Spread
`Spectrum
`Systems.
`Dissertation, University of Southern California, Los
`Angeles, CA, USA,, p240, 1982
`J. linatti, Matched Filter Code Acquisition Employing
`a Median Filter in Direct Sequence Spread-Spectrum
`Systems With Jamming. Acta Universitatis Ouluensis
`C102, Doctoral Thesis. University of Oulu, Oulu,
`
`Finland, pp54-65, 1997
`J.
`linati, DS Code Acquisition in Slowly Fading
`Multi-path Channel. proc. of IEEE VTC’2000 Fall,
`Boston, Ma USA, pp2408-2413, Sept. 24-28, 2000
`Andrew J. Viterbi, CDMA Principles of Spread
`
`2.
`
`ua
`
`4.
`
`$5.
`
`four
`chips are needed for FHT. Each FHT unit has
`accumulators with the bit width 10, 11,12 and 13. The
`
`streamline operation is used here.
`6. ACCUMULATOR
`
`To obtain the high process gain, we accumulate the FHT
`output 32 time in ACCUMULATORmodule.
`7. SQUARE
`
`Then accumulator values from 1I/Q branches are summed
`in this module.
`8. COMPARE
`
`The last module is COMPARE. Thecorrelation results
`
`of 16 signature are compared with the threshold that is set
`by higher layer successively. And the decision variable is
`obtained finally.
`
`VI. Conclusion
`
`investigated the PRACH
`this paper, we have
`In
`acquisition of WCDMA. Considering both acquisition
`performance and hardware scale, we proposed the novel
`fast acquisition scheme that based on FHT. The simulation
`ant test results are that the correct detection probability
`
`with E,/N, =7dB doesn’t
`
`less than 95% andthe single
`
`acquisition time is less than 5120chips.
`
`6.
`
`Spectrum Communication, Louay, 1995
`R.R.Rick and L.B.Milstein, Parallel Acquisition in
`Mobile
`DS-CDMA
`Systems.
`IEEE
`Trans.
`Communications, Vol.45, pp1466-1476, Nov. 1997
`7. W, Zhuang, Non-coherent Hybrid Parallel PN Code
`Acquisition for CDMA Mobile Communications.
`IEEE Trans. Vehicular Technology, Vol.45, pp643-
`656, Nov. 1996
`
`8.
`
`9.
`
`3G TS 25.2, V3.2.0 (2000-03): Physical channels
`and mapping of transport channels onto physical
`channels (FDD)
`
`J. linati, On the Threshold Setting Principles in Code
`Acquisition of DS-SS Signals.
`IEEE JSAC, Vol.18,
`pp62-72. January 2000
`
`10. Polydoros and C.L.Weber, A Uniforied Approach to
`Serial Search Spread-Spectrum Code Acquisition-Part
`] &Il, IEEEE Trans. Communication, Vol.COM-32,
`
`pp542-561, May 1984
`Ii. ETSI TR 101, 102, UMTS
`
`1705
`
`5
`
`