Where are the Relay Capacity Gains in Cellular Systems?

Size: px

Start display at page:

Download "Where are the Relay Capacity Gains in Cellular Systems?"

Della Fields
5 years ago
Views:

Austin Wireless Networking and Communications Group http://www.ece.utexas.

1 Where are the Relay Capacity Gains in Cellular Systems? Robert W. Heath Jr. Steven Peters, Kien Truong, and Ali Yazdan-Panah The University of Texas at Austin Wireless Networking and Communications Group * Funded by a gift from Huawei The Wireless Networking and Communications Group

2 Introduction to Relays Relays assist communication May be fixed (infrastructure), mobile (bus) or cooperative (other users) Single antenna, multiple antennas, multiple s, multiple users Main purpose of s Coverage (large-scale effects), diversity (small-scale) [?], and capacity [??]

3 Types of Relays st time-slot nd time-slot One-way full duplex One-way half duplex Two-way half duplex One-way multi-hop

4 Relay Operation r r = γr Amplify-and-forward Relay scales RX signal r E{ r } P r Abstract DeMOD Detect MOD r Compress r = Q (r) E{ r } P r Abstract r E{ r } P r Abstract Seems practical; less interest in standards Decode-and-forward Relay decodes RX signal Very practical Compress-and-forward Relay compresses RX sig Receiver decodes with partial information Less industry interest

direct transmission d 0 d R = As, bps/hz (DF is the best) R = As, bps/hz (CF is the best) R = log(

5 Expected Gains of Relays d -d Linear system: full-duplex, no fading & same transmit powers [KraEtAl05] R R direct = log( + kp) log( + P ) log(k) R as P The maximum capacity gain over the direct transmission d 0 d R = As, bps/hz (DF is the best) R = As, bps/hz (CF is the best) R = log( + α ) α d =0.5 α [, 5) If, bps/hz path-loss exponent Capacity gains of s are modest (there is diversity as well) 5

6 Cellular Systems Interference Limited Lots of interference Relay deployments 7% indoor coverage improv. [DoppEtAl08] 0% improvement in outage SINR w/ fast PC [SreEtAl0] Better uplnk coverage, 0% rate improvement GPP [IrmDie08] 6% rate improvement downlink LTE-A [LinEtAl09] With interference, capacity gains seem to reduce further 6

7 Purpose of this Talk () Discuss performance of different ing strategies () Suggest some solutions to make s better Compare with no & coordinated transmission (via DPC) 7

8 Simulation Framework Interfering Sectors Channel IEEE 80.6j Type E channel model 6 dbm base station transmit power 7 dbm transmit power dbm mobile transmit power Configuration cells, 6 sectors per cell, antenna UE Compute throughput on ring Switch from direct to link as move from BTS [ring based] 8

9 Relay Performance Downlink (DL) Uplink (UL) user achievable rate (bps/hz).5.5 Direct transmission BS Coordination Single antenna user achievable rate (bps/hz) 5 Direct transmission BS Coordination MS distance from BS (m) BS-RS interference limited MS distance to BS (m) BS-RS interference limited Half duplex, ignore direct link, DF, optimum time sharing No gain from a single (ouch!!!) 9

Add Multiple Antennas leakage signals desired signal First phase: Receive MMSE RX receive filter Second phase: TX beamforming (BF) at s MRT (max. ratio trans.

10 Add Multiple Antennas leakage signals desired signal First phase: Receive MMSE RX receive filter Second phase: TX beamforming (BF) at s MRT (max. ratio trans.): maximizing desired signal power to its own user ZF (zero-forcing): minimizing sum of leakage powers to other users SLNR (signal-to-leakage-plus-noise ratio): balancing MRT & ZF (like MMSE) 0

11 One-Way Relay with Antennas 7 user achievable rate (bps/hz) Single antenna + Direct BS coordination MRT antennas + Direct ZF antennas + Direct SLNR antennas + Direct MRT antennas + Direct ZF antennas + Direct SLNR antennas + Direct MRT antennas + Direct ZF antennas + Direct SLNR antennas + Direct MS distance from BS (m) Downlink (DL) Uplink (UL) Using multiple antennas at s provide significant gains user achievable rate (bps/hz) 6 5 Single antenna + Direct BS coordination MRT antennas + Direct ZF antennas + Direct SLNR antennas + Direct MRT antennas + Direct ZF antennas + Direct SLNR antennas + Direct MRT antennas + Direct ZF antennas + Direct SLNR antennas + Direct MS distance from BS (m) Better than direct trans./single-ant. s from half of cell radius to cell-edge Interference cancellation is good only for DL cell-edge users Enhanced functionality reduces gap to full coordination

12 Shared Relaying Multi-antenna placed at cell intersection [PetEtAl09] Relay shared among multiple base stations Inter-cell interference removed through MU-MIMO techniques Performs a decode and forward operation

13 Shared Relay Performance user achievable rate (bps/hz) Antenna Shared Relay + Direct Antenna Shared Relay + Direct 5 Antenna Shared Relay + Direct 6 Antenna Shared Relay + Direct BS Coordination Direct transmission MS distance to BS (m) user achievable rate (bps/hz) MS distance from BS Antenna Shared Relay + Direct Antenna Shared Relay + Direct 5 Antenna Shared Relay + Direct 6 Antenna Shared Relay + Direct BS Coordination Direct transmission Downlink Uplink Enhanced functionality reduces gap to full coordination

Two-Way Shared Relay at the signal level to construct

phase I Concept +γ γ at the signal level to construct t =

at the signal level to iconstruct i i MS + γ γ γ γ, i =,,.

Next, to spatially separate thesibs-ms pairs, the assigns

The transmitted vector from the is t = P Pr Wt, where Ww

ther assigns r i i beamforming Next, vectors toeach

assigns vectors wi to unique beamforming T with tr(ww

Wt, where W = [w, w th BS is T and P is signal in first

channels, = [t, t, tthe the total average power from the

Assuming ],received r with tr(ww ) =, t = [t, t, t ]T, and

in the channels, the received signal in first sector of

received ysignal intr first of the i BS is i = hi i, yi =

Similarly, atyithe = hith i tr + ni, phase II phase III

Similarly, at the ith MS th zi = at githe tr i+ vms where

Similarly, i, Cellular Topology zi = gi t + v i, (9) where

Viewing these zir= signals gi tr +invi,corresponding pairs

that i = (0, wherevector vi dcn Viewing these signals in

signals in corresponding pairs we define the vector di =

14 Two-Way Shared Relay at the signal level to construct inter-sector interf. phase I Concept +γ γ at the signal level to construct t = s + x, γ, i =,,BS. at the signal level to iconstruct i i MS + γ γ γ γ, i =,,. RN (shared) ti = si +x+i γ, (7) other MS interf. Next, to spatially separate thesibs-ms pairs, the assigns ti = + xi, unique γ,beamforming i =,,. vectors wi to ea t. The transmitted vector from the is t = P Pr Wt, where Ww =i[w, w, w w t = i to spatially separate the BS-MS pairs, ther assigns r i i beamforming Next, vectors toeach i=unique Next, to spatially separate the BS-MS pairs, the assigns vectors wi to unique beamforming T with tr(ww vector ) =, tfrom = [tthe, t ],isand the from the Assum average, where, t ti. The transmitted tr P=r is P Pr Wt W =terminal. [w, w w, w ] r total i ti= power i= ti. The transmitted vector from the is tr = Pr i= wi ti = Pr Wt, where W = [w, w th BS is T and P is signal in first sector of the i with reciprocity tr(ww ) in = the, t channels, = [t, t, tthe the total average power from the terminal. Assuming ],received r with tr(ww ) =, t = [t, t, t ]T, and Pr is the total average power from the terminal. Assu reciprocity in the channels, the received signal in first sector of the ith BS isth + nsector reciprocity in the channels, the received ysignal intr first of the i BS is i = hi i, yi = h i tr + n i,ms where ni CN (0, N0 ) is AWGN. Similarly, atyithe = hith i tr + ni, phase II phase III (8) where ni CN (0, N0 ) is AWGN. Similarly, at the ith MS th zi = at githe tr i+ vms where ni CN (0, N0 ) is AWGN. Similarly, i, Cellular Topology zi = gi t + v i, (9) where vi CN (0, N0 ) is AWGN. Viewing these zir= signals gi tr +invi,corresponding pairs we define the [yi,nz0i ])T issoawgn. that i = (0, wherevector vi dcn Viewing these signals in corresponding pairs we define the where vi CN (0, N0 ) is AWGN. Viewing these signals in corresponding pairs we define the vector di = [yi, zi]t so that di inter-sector = [hi gi ] trinterference + [ni, vi ]T Phase I: Relay decodes under vector dbs [yi, zi ]T so that i = signals T di = [h g ] t + [n, v ] i i i i R F Wt = rms i [ni, vi ]T = P[h ]+ trn+ Phase II: Relay decodes MS signals underd interference iother ii gi Wt + n = =Pr FiP i Fi wj tj + ni, ( rf irw iitwt i + +P ri P F n = Phase III: UL/DL power control + spatialorthogonalization block diag. via j=i i ti + Pr Fi w = Pr Fi w (0) j tj + ni, = P F w t + P F w t + n, is a composite BS-MSr channel i i i r ith cell i and j j n i CN (0, N I ). To enfo where Fi =of[hul = i the i gi i 0 A single superposition +] DL signals is transmittedjfor by j=i i.e. cancel interference other pairs, follow [hi gi ] isina (0), composite BS-MSthechannel for thefrom ith cell andbs-ms n CN (0, Nwe. Tothe enforce wherespatial Fi = separation 0 I )set (0, N0 I ). To en where Fi = [hi gi ] is a composite BS-MS channel for the ith celli and ni CN constraint on in the(0), beamforming Fi wj = 0 = ibs-ms. By defining the set the M following matrix, j spatial separation i.e. cancelvectors the interference from other pairs, we F i spatial separation in (0), i.e. cancel the interference from other BS-MS pairs, we set the foll

15 Two-Way Shared Relay Uplink user achievable rate [uplink] (bps/hz) Twoway + Direct Direct BS Coordination Break Point (50m) R UL (γ) = min{r(ii) UL,R(III) UL } in presence of MS interference w/ block diagonalization MS distance from BS (m) Performance Performance is reduced due to inter-sector and inter cell MS interference Direct transmission is preferred unless MS closer to shared Optimal time-share with phase III is also assumed (not shown in equation) 5

16 Downlink Comparison Area Spectral Efficiency (ASE) user achievable rate (bps/hz) Single antenna + Direct BS coordination One way SLNR (x) ants + Direct One way SLNR (x) ants + Direct One way shared ants + Direct One way shared 6 ants + Direct Two way shared 6 ants+ Direct MS distance from BS (m) Configurations One-way shared is good for cell-edge users One-way with multiple antennas help more-inner users Two-way shared does not help DL transmission ASE (bps/hz/km ) Single-antenna + direct 6.99 BS coordination 0.0 One-way SLNR (x) ants + Direct 8.5 One-way SLNR (x) ants + Direct.68 One-way shared ants + Direct 0. One-way shared 6 ants + Direct. Two-way shared 6 ants + Direct

17 Uplink Comparison user achievable rate (bps/hz) Single antenna + Direct BS coordination One way SLNR (x) ants + Direct One way SLNR (x) ants + Direct One way shared ants + Direct One way shared 6 ants + Direct Two way shared 6 ants+ Direct MS distance from BS (m) Area Spectral Efficiency (ASE) Configurations ASE (bps/hz/km ) Single-antenna + direct.78 BS coordination 0.6 One-way SLNR (x) ants + Direct 6. One-way SLNR (x) ants + Direct 9.08 One-way shared ants + Direct.96 One-way shared 6 ants + Direct 5.6 Two-way shared 6 ants + Direct 7.58 Two-way shared is good for cell-edge users One-way with multiple antennas help more-inner users One-way shared does not help UL transmission 7

18 Uplink / Downlink Sum Comparison user achievable rate (bps/hz) Single antenna + Direct BS coordination One way SLNR (x) ants + Direct One-way SLNR (x) ants + Direct One way shared ants + Direct One way shared 6 ants + Direct Two way shared 6 ants+ Direct Area Spectral Efficiency (ASE) Configurations ASE (bps/hz/km ) Single-antenna + direct 5.89 BS coordination 5. One-way SLNR (x) ants + Direct 7.8 One-way SLNR (x) ants + Direct 0.88 One-way shared ants + Direct 7.6 One-way shared 6 ants + Direct 8.0 Two-way shared 6 ants + Direct MS distance from BS (m) Two-way shared is not as good as one-way shared 8

19 Where are the Capacity Gains? Not here: Relays that neglect interference They arguably suck even without interference Here: Relays that deal with interference Multiple antennas improve mid-range cell performance Multiple antenna shared improves edge of cell performance Combination of strategies seems very attractive Future work: Two-way, selection, power control 9

20 References [DoppEtAl08] K. Doppler, C. Wijting, and K. Valkealahti, On the Benefits of Relays in a Metropolitan Area Network, VTC 008. [SreEtAl0] V. Sreng, H. Yanikomeroglu, and. D. Falconer, Coverage enhancement through two-hop ing in cellular radio systems, WNCC 00. [IrmDie08] R. Irmer and F. Diehm, On coverage and capacity of ing in LTEadvanced in example deployments," PIMRC 008. [LinEtAl09] Huang Lin, Daqing Gu, Wenbo Wang, Hongwen Yang, "Capacity analysis of dedicated fixed and mobile in LTE-Advanced cellular networks," ICCTA 009. [PetEtAl09] S. W. Peters, A. Y. Panah, K. T. Truong, and R. W. Heath, Jr., ``Relaying Architectures for GPP LTE-Advanced,'' EURASIP Journal on Advances in Signal Processing, special issue on GPP LTE and LTE Advanced, vol. 009, Article ID 68787, pages, doi:0.55/009/68787, 009. [KraEtAl05] G. Kramer, M. Gastpar, and P. Gupta, Cooperative strategies and capacity theorems for networks, IEEE Trans. Info. Theory, no. 9, vol. 5, pp , Sep

Uplink and Downlink Rate Analysis of a Full-Duplex C-RAN with Radio Remote Head Association

Uplink and Downlink Rate Analysis of a Full-Duplex C-RAN with Radio Remote Head Association Mohammadali Mohammadi 1, Himal A. Suraweera 2, and Chintha Tellambura 3 1 Faculty of Engineering, Shahrekord