Forschungszentrum Telekommunikation Wien OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) T. Zemen April 24, 2008
Outline Part I - OFDMA and SC/FDMA basics Multipath propagation Orthogonal frequency division multiplexing (OFDM) Soft frequency reuse Peak-to-average power ratio (PAPR) Single carrier/frequency division multiple access (SC/FDMA) Part II - Time-variant channel estimation for OFDM Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 2/37
References Motorola, Long Term Evolution (LTE): Overview of LTE Air-Interface, Technical White Paper, 2007 Rohde&Schwarz, UMTS Long Term Evolution (LTE) Technology Introduction, 2007. Freescale, Overview of the 3GPP Long Term Evolution Physical Layer, 2007. 3GPP TS 36.2111, Physical Channels and Modulation (Rel. 8), March, 2008. H. G. Myung, J. Lim, and D. J. Goodman, Single Carrier FDMA for Uplink Wireless Transmission, IEEE Vehicular Technology Magazin, pp. 30-38, September, 2006. U. Sorger, I. De Broeck, and M. Schnell, Interleaved FDMA - A New Spread Spectrum Multiple-Access Scheme, Proc. IEEE ICC 98, Atlanta, GA, pp-1013-1017, June, 1998. T. Zemen and C. F. Mecklenbräuker, Time-Variant Channel Estimation using Discrete Prolate Spheroidal Sequences, IEEE Transactions on Signal Processing, vol. 53, no. 9, September 2005, pp. 3597-3607. Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 3/37
Time-Variant Multipath Propagation scatterer η 1 e j2πf 1t δ(t τ 1 ) v velocity l path η 0 e j2πf 0t δ(t τ 0 ) η l τ l attenuation time delay user v receiver η 2 e j2πf2t δ(t τ 2 ) f l L Doppler shift number of paths scatterer Time-variant channel impulse response h(t, τ) = L 1 l=0 η l e j2πf lt δ(t τ l ) Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 4/37
OFDM Fundamentals (I) Single carrier versus multi carrier f single carrier f multi carrier d[0] d[1] d[2] d[0] d[1] d[2] d[3] d[4] d[5] d[6] d[7] d[3] d[4] d[5] d[6] d[7] t 0 T C 0 NT C t 1/T C Chip rate N Number of subcarriers d[0]... d[7] Data symbols Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 5/37
OFDM Fundamentals (II) Orthogonal subcarriers magnitude f 1...... 0 f f q f q+1 f q = q/(nt C ) q {0,..., N 1} Subcarrier index Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 6/37
Efficiently implementable by means of an inverse discrete Fourier transform. Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 7/37 OFDM Fundamentals (III) Processing steps subcarriers f 1 symbols (BPSK) modulated subcarriers t *(+1) t 2*f 1 t *(-1) t + t 3*f 1 t *(+1) t
OFDM Fundamentals (IV) Cyclic prefix insertion t T S T S OFDM symbol duration. A copy of the signal tail (length T G ) is inserted at the beginning of each OFDM symbol. Absorbs multipath components. Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 8/37
OFDM Fundamentals (V) OFDM time frequency representation Symbols FFT Guard Intervals 5 MHz Bandwidth Sub-carriers Frequency Time Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 9/37
OFDM System Design No inter-symbol interference: Guard interval larger than the delay spread T D GT C = T G > T D Spectral efficiency: Symbol duration much larger than delay spread NT C = T S T D Inter-carrier interference: Symbol rate much higher than Doppler shift f D 1/T S f D G cyclic prefix length in number of chips N number of subcarriers Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 10/37
Receiver Side Processing Drop cyclic prefix and perform DFT Channel partitioned in N parallel frequency flat channels Simple equalization - complexity grows with N log(n) g[0] n[0] d[0]... g[ q] n[ q] y[0]... d[ q] y[ q]...... q subcarrier index d data symbol g subcarrier channel coefficient n additive noise y received symbol d[ N-1]...... y[ N-1] Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 11/37
OFDMA (I) Orthogonal time-frequency grid a subcarrier frequency an OFDM symbol time Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 12/37
OFDMA (II) Time division multiple access (TDMA) frequency diversity frequency user 1 user 2 user 3 user 4 user 5 user 6 time Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 13/37
OFDMA (II) Frequency division multiple access (FDMA) frequency user 1 user 2 user 3 user 4 user 5 user 6 time time diversity Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 13/37
LTE Parameters (Downlink) Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 14/37
LTE Resource Grid (Downlink) Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 15/37
LTE Resource Blocks (Downlink) Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 16/37
OFDMA (III) Time-variant frequency-selective channel frequency frequency diversity time time diversity Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 17/37
OFDMA (III) Time-frequency pattern frequency frequency diversity user 1 user 2 user 3 user 4 time time diversity Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 17/37
Soft Frequency Reuse (I) At cell boundaries the signal to interference ratio (SIR) is approx. 0 db due to frequency reuse 1 DS-CDMA uses soft handover at the cell boundary OFDM based air interface allows soft frequency reuse for users at the cell boundary Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 18/37
Soft Frequency Reuse (II) power density BS 2 MS 22 BS 1 Power density MS 12 MS 11 MS 21 MS 31 Power density MS 32 subcarrier Subcarriers subcarrier BS 3 Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 19/37
LTE Uplink Power consumption in user equipment (UE) terminals is limited by battery OFDM requires large dynamic range due to high peak to average power ratio (PAPR) Linear power amplifiers with wide dynamic range have bad efficiency Single carrier/frequecy division multiple access (SC/FDMA) used for the uplink in LTE Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 20/37
Comparison OFDMA vs. SC/FDMA Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 21/37
SC/FDMA Distributed vs. Localized Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 22/37
SC/FDMA Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 23/37
Processing Steps Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 24/37
PAPR comparison Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 25/37
Conclusions - Part I OFDM enables wireless communication in frequency selective channels Channel equalization in the frequency domain Low complexity implementation - fast Fourier transform (FFT) algorithms Time and frequency diversity can be exploited SC/FDMA is OFDM with precoding PAPR of SC/FDMA is 2dB smaller compared to OFDMA Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 26/37
Outline Part I - OFDMA and SC/FDMA Basics Part II - Time-variant channel estimation for OFDM Signal model Discrete prolate spheroidal sequences Reduced-rank channel estimation Mean square error (MSE) sensitivity to velocity of user equipment Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 27/37
Time-Variant Channel Estimation for OFDM Signal model for a single flat-fading subcarrier q {0,..., N 1} y m,q = g m,q d m,q + z m,q y m,q z m,q d m,q C, received symbol on subcarrier q at discrete time m {0,..., M 1} C, additive noise C data symbol Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 28/37
LTE Pilot Pattern Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 29/37
Reduced-Rank Channel Description Second order statistic is unknown Doppler bandwidth upper bounded by ν Dmax Estimation per frequency-flat subcarrier q {0,..., N 1} Basis Expansion D 1 g m,q γ i,q u i,m for m {0,..., M 1}, D M i=0 g m,q u i,m γ i,q channel coefficient for subcarrier q at time m basis function expansion coefficient Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 30/37
Slepian Basis Expansion D. Slepian et al. (1961,1978) asked, which sequences have 1 maximum time concentration λ = M 1 u m 2 m=0 u m 2 m= 2 while being bandlimited in [ ν Dmax, ν Dmax ] M 1 l=0 sin(2πν Dmax (l m)) u i,l = λ i u i,m π(l m) Discrete prolate spheroidal (DPS) sequences u i [m] Doubly orthogonal on {,..., } and {0,..., M 1} Subspace has dimension D = 2ν Dmax M + 1 Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 31/37
Slepian Sequences 0.15 0.1 0.05 0 0.05 0.1 0.15 u 0 [m] u 1 [m] u 2 [m] u 3 [m] u 4 [m] M = 256 block length 1/T S = 49 ks 1 symbol rate v max = 100 km/h velocity f C = 2 GHz carrier D = 5 dimensions 0 50 100 150 200 250 m Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 32/37
Time-Variant Channel Estimation Signal model for a single subcarrier q y m,q = g m,q d m,q + z m,q y m,q z m,q γ i,q C, received symbol on subcarrier q at discrete time m C, additive noise C, basis expansion coefficient Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 33/37
Time-Variant Channel Estimation Signal model for a single subcarrier q y m,q z m,q γ i,q D 1 y m,q = u i,m γ i,q d m,q + z m,q i=0 C, received symbol on subcarrier q at discrete time m C, additive noise C, basis expansion coefficient Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 33/37
Pilot Based Channel Estimation (I) D 1 y m,q = u i,m γ i,q d m,q + z m,q i=0 Time multiplexed pilot and data symbols, d m = b m + p m (pilots are known at the receiver side). b m {±1 ± j}/ 2 for m / P and b m = 0 for m P p m = {±1 ± j}/ 2 for m P and p m = 0 for m / P { P = i M J + M } i {0,..., J 1} 2J J pilot symbols Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 34/37
Pilot Based Channel Estimation (II)... omitting q Basis function vector f m = Coefficient estimates ˆγ = G 1 u 0,m. u D 1,m m P R D, y m p mf m where ˆγ = [ˆγ 0,..., ˆγ (D 1) ] T and the correlation matrix G = l P f l f H l. Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 35/37
MSE Sensitivity to User Velocity MSE M = 1 M M 1 E { g m g 2} m m=0 Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 36/37
Conclusions - Part II Detailed channel information at receiver side difficult to acquire Discrete prolate spheroidal sequences describe the subspace of band-limited sequences Reduced-rank channel description using DPS sequences allows for simple UE algorithm Mean-square error (MSE) is practically independent of user velocity Thomas Zemen OFDMA/SC-FDMA Basics for 3GPP LTE (E-UTRA) April 24, 2008 37/37