`for OFDM Cellular Systems
`
`Kwang Soon Kim
`Electronics and Telecommunications
`Research Institute (ETRI)
`161 Gajeong-Dong, Yuseong-Gu,
`Daejeon 305-350, Korea
`Tel: +82-42-860-6702
`Fax: +82-42-869-6732
`E-mail: kwangsoon@etri.re.kr
`
`KyungHi Chang
`Grad. School of Information
`Technology and Telecommunication,
`Inha University
`253 Yonghyun-Dong, Nam-Gu,
`Incheon 402-751, Korea
`Tel: +82-32-860-8422
`Email: khchang@inha.ac.kr
`
`Sung Woong Kim
`and Yong Soo Cho
`Dept. Electronic Engineering,
`Chung-Ang University,
`221 HukSuk-Dong, Dongjak-Gu,
`Seoul 156-756, Korea
`Tel: +82-2-820-5299
`E-mail : yscho@cau.ac.kr
`
`new synchronization and cell searching technique for OFDM-
`based cellular systems. In this paper, a new preamble structure,
`including a synchronization field (S-field) and a cell searching
`field (C-field), is proposed. An efficient algorithm for downlink
`synchronization and cell searching using the preamble is also
`proposed and verified by extensive computer simulations.
`
`II. PREAMBLE DESIGN
`
`In this section, a novel preamble structure for an OFDM
`based cellular system is proposed. The downlink frame struc-
`ture considered in this paper is shown in Fig. 1. A preamble,
`with length Tp, is located at the beginning of the frame and is
`followed by a number of data slots, where pilot symbols are
`well spread in time and frequency. The proposed preamble
`is shown in Fig. 2. It is comprised of two fields, denoted
`as S-field and C-field. The S-field is designed for time and
`frequency synchronization and has its length, Tps, equal to the
`OFDM symbol duration, Ts. The S-field signal is composed
`(cid:1)
`(cid:1)
`symbol. The IS
`symbol is the
`of one S symbol and one IS
`first TG-length part of the π-phase rotated version of the S
`symbol, and the S symbol is comprised of NSsym repetitive Sa
`symbols. One good example for the preamble S-field signal,
`PS(t), is
`
`Abstract— In this paper, a novel preamble structure, including
`a synchronization field (S-field) and a cell searching field (C-field),
`is proposed. An efficient algorithm for downlink synchronization
`and cell searching using the preamble is also proposed. The
`synchronization process includes initial symbol timing estimation
`using continuously or at least periodically transmitted down-
`link signal, frame detection, fine symbol timing estimation and
`frequency offset estimation using the preamble S-field, and cell
`identification using the preamble C-field. From the simulation
`results, it is shown that the proposed preamble and cell searching
`algorithm works well even in bad cellular environment.
`
`I. INTRODUCTION
`
`frequency division multiplexing
`Recently, orthogonal
`(OFDM) has been widely accepted as the most promising
`radio transmission technology for the next generation wireless
`systems due to its advantages such as the robustness to
`multipath fading, granular resource allocation capability, and
`no intracell
`interference. Among the conventional OFDM-
`based wireless systems, digital audio broadcasting (DAB),
`digital video broadcasting, IEEE 802.11a, and Hiperlan/2 are
`well-known [1]-[5]. For cellular systems, it is one of the most
`important requirement to provide robust synchronization and
`cell searching capability. For example, the wideband code divi-
`sion multiple access (WCDMA) system provides a hierarchical
`three-step cell search using the primary synchronization code
`(PSC), the secondary synchronization code (SSC), and the
`common pilot channel (CPICH) [6]. During the initial cell
`search, a mobile station can obtain the frame timing and the
`scrambling code number of the best cell site.
`In DAB systems, null symbols and phase reference symbols
`are used for synchronization such as frame detection, symbol
`timing, integer and fractional part of frequency offset. On the
`other hand, a short preamble and a long preamble are used
`for burst synchronization in IEEE 802.11a and Hiperlan/2
`systems, such as frame detection, symbol timing, coarse and
`fine frequency offset, channel estimation [7]. However, these
`schemes are not appropriate for a cellular system since they
`cannot discriminate signals from different cells unless their
`carrier frequencies are different. Thus, it is required to devise a
`
`
`(cid:4)NF −1
`−(cid:4)NF −1
`0 ≤ t < Td,
`
`g(k)ϕk(t)
`g(k)ϕk(t − Td) Td ≤ t < Ts,
`0
`otherwise,
`−j2π(k−NF /2)t/Td, Td = Ts − TG, NF is the
`where ϕk(t) =e
`(cid:5)
`number of FFT points,
`
`PS(t) =
`
`k=0
`
`k=0
`
`g(k) =
`
`µ(i) k = iNSsym,
`0
`otherwise,
`
`(1)
`
`(2)
`
`and µ(i) is a pseudo-noise sequence, such as the m-sequence.
`Among many possible S-field signals given by (1) and (2),
`the signal with a low peak-to-average power ratio and good
`correlation property was selected as the preamble. Note that,
`although the S-field is similar to the preamble used in the
`Hiperlan/2 [2], the proposed preamble, designed and optimized
`
`0-7803-7954-3/03/$17.00 ©2003 IEEE.
`
`2471
`
`Authorized licensed use limited to: Sterne Kessler Goldstein Fox. Downloaded on July 13,2023 at 21:12:20 UTC from IEEE Xplore. Restrictions apply.
`
`VWGoA EX1039
`VWGoA V. Neo Wireless
`IPR2022-01539
`
`
`
`TP
`
`P
`
`Tframe
`
`Data
`Slot
`
`Data
`Slot
`
`Data
`Slot
`
`Data
`Slot
`
`Fig. 1. An abstract downlink frame structure.
`
`for a cellular system, is very different from those used in the
`Hiperlan/2 in terms of usage and synchronization algorithm.
`The C-field is designed for cell search and has its length
`NcTs. The C-field signal of the mth cell, P m
`C (t), is defined
`as
`P m
`C (t)
`
`
`
`
`=
`
`(cid:4)NF −1
`(cid:4)NF −1
`n (k)ϕk(t + Td − TG),
`cm
`n (k)ϕk(t − TG),
`cm
`0,
`
`k=0
`
`k=0
`
`0 ≤ t − nTs < TG,
`TG ≤ t − nTs < Ts,
`otherwise,
`
`(3)
`
`where TG is the guard interval, n is in the range between
`0 and Nc − 1, and cm
`n (k) is the frequency domain sig-
`nal of the nth symbol of the C-field in the mth cell. Let
`F = {f0, f1, · · · , fNu−1} be the set of used subcarrier set,
`where Nu represents the number of used subcarrier, and
`S = {s0, s1, · · · , sP−1} be a disjoint partition of F such
`that, for any 0 ≤ i (cid:2)= j < P , si ∈ F , si ∩ sj = ∅, and
`∪P−1
`i=0 si = F . For theith partition si, we divide it intos i,p
`and si,c such that si,p ∪ si,c = si and si,p ∩ si,c = ∅. Let
`Ψi = {ψi,0, ψi,1 · · · , ψi,Qi−1} be a set of complex sequences
`with length equal to the cardinality of si,c, Ji, and with good
`auto- and cross-correlation property. Further, we define ¯ψ as a
`complex sequence used for known pilot symbol pattern, whose
`length is equal to the cardinality of si,p. Then, cm
`n (k) is defined
`as
`
`
`ψpn,qn(j) k = spn,c(j),
`¯ψ(j)
`k = spn,p(j),
`0
`otherwise,
`where ψpn,qn(j), ¯ψ(j), spn,p(j), and spn,c(j) are the jth
`element of ψpn,qn, ¯ψ, spn,p, and spn,c, respectively,
`pn−1(cid:6)
`P−1(cid:6)
`
`cm
`n (k) =
`
`mn =
`
`Qi + qn,
`
`i=0
`
`M =
`
`Qi,
`
`(4)
`
`(5)
`
`Fig. 2. The proposed preamble structure.
`
`Fig. 3. The proposed synchronization process.
`
`III. SYNCHRONIZATION ALGORITHM
`
`In Fig. 3, the synchronization process proposed in this paper
`is shown. In the conventional OFDM-based systems, such as
`wireless LAN, all initial synchronization process, including
`signal detection, is done by using the preamble. However, in
`an OFDM-based cellular system, signal is transmitted contin-
`uously or, at least, periodically due to common pilot symbols
`and common channels used for broadcasting. In this situation,
`an initial symbol timing and initial frequency offset can be
`obtained by using the cyclic prefix (CP). After achieving initial
`synchronization, we estimate the frame timing and fine symbol
`timing by using the preamble S-field.
`
`and
`
`m =
`
`i=0
`
`Nc−1(cid:6)
`
`mnM Nc−n−1.
`
`(6)
`
`n=0
`Here, we can see that, with the proposed C-field, M Nc dif-
`ferent cells can be discriminated. As an example, let Nc = 2,
`P = 8, Qi = ¯Q = 8. Then, the number of different cells
`is M Nc = 642 = 4096, which is large enough for a cellular
`system.
`
`0-7803-7954-3/03/$17.00 ©2003 IEEE.
`
`2472
`
`Authorized licensed use limited to: Sterne Kessler Goldstein Fox. Downloaded on July 13,2023 at 21:12:20 UTC from IEEE Xplore. Restrictions apply.
`
`
`
`A. Initial symbol timing and frequency offset estimation
`Let the sampled received signal be y(n). Then, the initial
`symbol timing, τinit, is estimated as
`
`(cid:7)(cid:7)(cid:7)(cid:7)(cid:7)(cid:7) 1
`
`NCP
`
`Ninit−1(cid:6)
`
`j=0
`
`τinit = max
`n
`
`(cid:7)(cid:7)(cid:7)(cid:7)(cid:7)(cid:7) ,
`
`z(n + jNs)
`
`(7)
`
`where
`
`NCP −1(cid:6)
`
`∗
`
`y
`
`z(n) =
`
`(n + r)y(n + r + NF ).
`
`(8)
`
`r=0
`Also, NCP , Ns, and Ninit are the number of samples in the
`guard interval, the number of samples in an OFDM symbol,
`and the number of OFDM symbols used in the initial symbol
`timing estimation, respectively. In addition, we can estimate
`the initial frequency offset, init, as
`
`
`Ninit−1(cid:6)
`
`j=0
`
` init =
`
`1
`2π
`
`arg
`
`
` .
`
`z(τinit + jNs)
`
`(9)
`
`Note that only fractional part of frequency offset can be
`estimated by (9).
`
`(10)
`
`B. Frame timing estimation
`After obtaining initial synchronization, we have Nf rame =
`Tf rame/Ts candidates for the frame timing. Here, we can
`utilize the property of the S-field where every OFDM symbol
`except
`the preamble S-field has a positive value of auto-
`correlation due to the cyclic prefix. Note that the correlation
`(cid:1)
`part in the S-field becomes
`between the Sa part and the IS
`negative. Then, the frame timing is estimated as
`τf (i) =aN s + τinit, 0 ≤ a < Nf rame
`τf rame = τf (i), if (cid:7){z(τf (a))} < 0,
`where (cid:7){x} is the real part of x.
`C. Fine symbol timing and coarse frequency offset estimation
`After obtaining the frame timing, we can assume that the
`starting point of the preamble S-field is around the estimated
`the frame timing τf rame. Then, we can estimate the fine
`symbol timing, τs, by taking cross-correlation between the
`received signals and the preamble S-field signal as
`
`(cid:7)(cid:7)(cid:7)(cid:7)(cid:7) ,
`
`(11)
`
`(r + Rs)y(n + r + Rs)
`
`∗S
`
`P
`
`(cid:7)(cid:7)(cid:7)(cid:7)(cid:7)R−1(cid:6)
`
`τs = max
`n
`
`r=0
`where PS(r) is the sampled signal of the preamble S-field,
`R is the number of samples used for the fine symbol timing
`estimation, and Rs is the starting point for accumulation. We
`can also estimate frequency offset, f , by using the repetitive
`property of the preamble S-field, as
`(cid:11)
` f = NSsym
`NCP −1(cid:6)
`2π
`· arg
`
`Fig. 4. Schematic for the cell identification.
`
`Note that the range of frequency offset estimated by (12) is
`[−NSsym/2, NSsym/2].
`IV. CELL IDENTIFICATION ALGORITHM
`In Fig. 4, the schematic for cell identification, proposed
`in this paper, is shown. Let Yn(k) be the frequency domain
`symbols at the kth sub-carrier in the nth OFDM symbol. Then,
`for the nth symbol of the preamble C-field, the sub-carrier
`partition pn and sequence number qn can be estimated by
`|Yn(k)|2
`
`(cid:6) k
`
`∈sp
`
`ˆpn = arg max
`p
`
`and
`
`where
`
`(cid:5)
`
`ˆqn =
`
`χp,q =
`
`arg maxq (cid:7) {χ ˆpn,q} , χ ˆpn,ˆqn > χth,
`Detectionf ails,
`otherwise,
`Jp−1(cid:6)
`
`∗p
`
`(sp,c(j))ψ
`
`,q(j),
`
`Yn(sp,c(j)) ˆH
`
`∗
`
`j=0
`
`(13)
`
`(14)
`
`(15)
`
`χth is the threshold for cell identification, and ˆH(k) is the
`estimated complex channel gain at the kth sub-carrier obtained
`from Yn(k), k ∈ s ˆpn,p, and the sequence ¯ψ. Then, the cell ID
`ˆpn−1(cid:6)
`is estimated as
`Nc−1(cid:6)
`
`ˆmn =
`
`Qi + ˆqn
`
`i=0
`
`ˆm =
`
`ˆmnM Nc−n−1.
`
`(16)
`
`(17)
`
`and
`
`n=0
`V. SIMULATION RESULTS
`The system parameters used in the simulation are given as
`follows:
`• carrier frequency : 2 GHz
`• bandwidth : 20 MHz
`• FFT size (NF ): 2048
`• data duration (Td) : 102.4 us
`• guard interval (TG) : 25.6 us
`• symbol duration (Ts) : 128 us
`• number of symbols used in initial
`(Ninit) : 3
`• number of preamble C-field symbol (Nc) : 1
`• number of partitions (P ) : 8
`• number of sequences in a partition ( ¯Q) : 1
`• channel model : ITU-R SISO model
`
`timing estimation
`
`(cid:12)
`
`∗
`
`y
`
`(τs + r)y(τs + r + NF F T /NSsym)
`
`.
`
`r=0
`
`(12)
`
`0-7803-7954-3/03/$17.00 ©2003 IEEE.
`
`2473
`
`Authorized licensed use limited to: Sterne Kessler Goldstein Fox. Downloaded on July 13,2023 at 21:12:20 UTC from IEEE Xplore. Restrictions apply.
`
`
`
`TABLE I
`JOINT INITIAL TIMING ESTIMATION AND FRAME DETECTION
`PERFORMANCE.
`
`Channel
`(Speed)
`
`SINR=0dB
`FA
`
`SINR=0dB
`DF
`
`SINR=5dB
`FA
`
`SINR=5dB
`DF
`
`AWGN
`
`< 2.0e-4
`
`< 2.0e-4
`
`< 2.0e-4
`
`< 2.0e-4
`
`Ped. B
`(3km/h)
`
`Veh. A
`(100km/h)
`
`1.28e-2
`
`< 2.0e-4
`
`2.0e-4
`
`< 2.0e-4
`
`6.68e-2
`
`4.0e-4
`
`7.4e-3
`
`< 2.0e-4
`
`Cell Scenario
`
`Cell 7
`
`Cell 2
`
`Cell 4
`
`Cell 1
`
`Cell 6
`
`Cell 0
`
`Cell 5
`
`Cell 3
`
`10 km
`
`: Mobile Station
`
`Fig. 5. Simulation scenario
`
`• path loss model : Hata model (COST-231)
`• cell radius : 10 km
`• cell scenario : see Fig. 5
`In table I, the joint performance of initial symbol timing
`estimation and frame detection is shown. The simulation
`results were obtained from 5000 trials under the assumption
`that inter-cell interference can be modelled as an additive white
`Gaussian noise (AWGN). Here, ’<2.0e-4’ signifies that no
`error is found in the 5000 trials. From table I, it is seen that
`the joint performance of the initial symbol timing estimation
`and frame detection degrades as the mobile speed increases.
`When the signal-to-interference-and-noise ratio (SINR) is 5dB
`and the mobile speed is 100km/h in the Vehicular A channel,
`we have the false alarm (FA) probability equal to 7.4e-3 and
`detection failure (DF) probability less than 2.0e-4.
`In Fig. 6, the performance of the fine symbol timing esti-
`mation is shown. In the simulation, the inter-cell interference
`is assumed to be AWGN and the normalized carrier frequency
`offset (CFO) is assumed to be in [−2.0, 2.0]. As can be
`seen from Fig. 6, the performance of the fine symbol timing
`estimation degrades as the maximum delay spread increases.
`At
`the SINR of 5dB and the mobile speed of 3km/h in
`the Pedestrian B channel,
`the probability that
`the timing
`estimation error less than 30 samples is 0.98 and the maximum
`timing error occurred in this simulation is 57 samples. In the
`Vehicular A channel, the probability that the timing estimation
`error less than 27 samples is 0.99 and the maximum timing
`
`Fig. 6. Fine timing estimation performance.
`
`Fig. 7. Frequency offset estimation performance.
`
`error occurred in this simulation is 50 samples.
`In Fig. 7, the overall performance on the frequency offset
`estimation is shown when the normalized CFO is set at 1.3.
`From Fig. 7, we can see that the performance degrades as the
`mobile speed increases. Also, the mean squared error (MSE)
`at the SINR of 2dB is lower than 1e-3 even in the Vehicular
`A channel with the mobile speed of 250km/h. Although not
`shown in this paper, we could observe that there is a negligible
`performance loss for up to 16-QAM symbol transmission if
`the MSE of the estimated frequency offset is below 1e-3. In
`the Pedestrian B channel with the mobile speed of 3km/h, we
`can achieve the MSE lower than 1e-4 at the SINR of 3dB or
`greater.
`
`0-7803-7954-3/03/$17.00 ©2003 IEEE.
`
`2474
`
`Authorized licensed use limited to: Sterne Kessler Goldstein Fox. Downloaded on July 13,2023 at 21:12:20 UTC from IEEE Xplore. Restrictions apply.
`
`
`
`When the mobile is at the cell boundary, the probability that
`cell ID 0 is not estimated is around 0.5. However, it does not
`necessarily mean that the cell identification procedure fails
`since the estimated cell ID with a stronger power is 1 in this
`situation. Thus, the cell identification in this case is regarded
`as successful. From Fig. 8, we can also see that the proposed
`preamble and cell identification algorithm works well even in
`fading channel at the SINR of 0 dB.
`
`VI. CONCLUSION
`In this paper, a preamble based synchronization and cell
`searching technique for OFDM cellular system was proposed.
`The preamble, composed of S-field and C-field, and the algo-
`rithm for synchronization and cell searching were proposed.
`From simulation results, we confirmed that
`the proposed
`approach could achieve very robust synchronization and cell
`searching performance even in bad cellular environments.
`Performance analysis on the proposed algorithms and the
`overall cell searching performance analysis in terms of mean
`acquisition time is our future work.
`
`REFERENCES
`[1] IEEE 802.11a, High-speed physical layer in the 5 GHz band, 1999.
`[2] ETSI BRAN TS 101 475, Broadband radio access networks(BRAN)
`HIPERLAN type 2: physical(PHY) layer, April 2000.
`[3] ETSI EN 300 401, Radio broadcasting systems: digital audio broadcast-
`ing (DAB) to mobile, portable and fixed receivers, September 2000.
`[4] ETSI EN 300 799, Digital video broadcasting (DVB); framing, structure,
`channel coding and modulation for digital terrestrial television, June
`1999.
`[5] IEEE 802.16ab-01/01r1, An Air Interface for Fixed Broadband Wireless
`Access Systems Part A: Systems between 2 and 11 GHz, July 2001.
`[6] 3GPP TS 25.125, Physical layer procedures (FDD), March 2000.
`[7] R.V. Nee and R. Prasad, OFDM Wireless multimedia communications,
`Artech House, 2000.
`
`Fig. 8. Cell identification performance.
`
`In Fig. 8, the performance of cell identification in various
`channel environment is shown. Here, the signal-to-noise ratio
`is set at 5dB and the distance between the base station in cell
`0 and the mobile station varies from 7km to 12km. As the
`distance varies, the SINR in 8-cell environment varies from
`1.53dB to 7.20dB. As can be seen from Fig. 8, the cell ID
`0 is estimated with a probability greater than 0.999 in the
`distance range between 7km and 9.5 km.
`
`0-7803-7954-3/03/$17.00 ©2003 IEEE.
`
`2475
`
`Authorized licensed use limited to: Sterne Kessler Goldstein Fox. Downloaded on July 13,2023 at 21:12:20 UTC from IEEE Xplore. Restrictions apply.
`
`