Coordinated and Distributed MIMO turning wireless networks on their heads? Gerard Borg 1 Coordinated and Distributed MIMO
Outline Orientation: Coordinated and distributed MIMO vs SISO Theory: Capacity and range of a CDM infrastructure network Simulations: Capacity and BER of a CDM infrastructure network Technical and protocol issues A cellular network comparison Other Architectures 2 Coordinated and Distributed MIMO
Current Wireless Networks Fundamentally Single Input Single Output (SISO)-like Systems (infrastructure = cellular) Characterised by one transmitting station at a time. One or many receiving stations. Examples: SISO, Single User-MIMO (Multiuser-MIMO). Issues Throughput limited by Shannon single channel capacity (unable to escape spectrum constraints) Range and coverage limited by antenna EIRP limit Colocated antennas at the transmitter and receiver limits the number of independent channels that can be support in SU-MIMO - low channel rank Often backhaul limited too. 3 Coordinated and Distributed MIMO
SISO Infrastructure Network Internet Remotes backbone Base bottlenecks 4 Coordinated and Distributed MIMO
SU-MIMO / MU-MIMO 5 Coordinated and Distributed MIMO
Coordinated and distributed MIMO Networks I More transmitters than receivers. MISO systems Examples: Sensor Networks, CO-MIMO cellular networks (still the same number of base stations?) Assumptions Bases and remotes all use the same spectrum. Uplink (all remotes to all bases) and downlink (all bases to all remotes) occur in consecutive time slots. Bases use matched channel precoding (i.e. the base array is a time reversal mirror - possibly the simplest precoding strategy. Bases stations have to be synchronised Remotes do not need to be synchronised 6 Coordinated and Distributed MIMO
Distributed Infrastructure Network II (Shidong Zhou et. al. (2003) 7 Coordinated and Distributed MIMO
Coordinated and distributed MIMO networks III: Characteristics Throughput limited by Shannon single channel capacity for each receiver (multiple bases allow reuse of spectrum for multiple receivers - akin to MU-MIMO). Cochannel interference limited Range and coverage no longer limited by antenna EIRP but by the number and placement of bases. For multiple distributed single antenna transmitters, array aperture is huge - there will always be a quasi-infinite channel rank and scalable net capacity Parallel backhaul? Main issues: Synchronisation, signal procesing overhead and information transport - turn out to be implementation issues 8 Coordinated and Distributed MIMO
SISO vs CDM Infrastructure Network IV 9 Coordinated and Distributed MIMO
Distributed Infrastructure Network V Internet Internet Internet Remotes High speed backbone Bases 10 Coordinated and Distributed MIMO
Theory 11 Coordinated and Distributed MIMO
Theory: Matched Channel Precoding (Time Reversal Precoding I) N base stations and M remote stations. ISI-free. Let x(t) be (QPSK) data symbols Signal transmitted from the n th base station, y n (t) = 1 M 1 M µ=0 h nµx µ (t) (1) Signal received at the m th remote station, z m (t) = N 1 n=0 h nm y n (t) + Ñ (2) 12 Coordinated and Distributed MIMO
Theory: Matched Channel Precoding (Time Reversal Precoding II) N base stations and M remote stations. Include ISI. Signal transmitted from the n th base station, y n (t ) = 1 M 1 M µ=0 T 1 t =0 h nµ (t + t )x µ (t ) (3) Signal received at the m th remote station, z m (t) = N 1 n=0 T 1 t =0 h nm (t t )y n (t ) + Ñ (4) 13 Coordinated and Distributed MIMO
Simple Capacity: M N: Interference limited: No ISI Each base launches x m (t) with power 1/M of EIRP At m th remote the wanted signal arrives with strength N/ M At m th remote the wanted signal arrives with power N 2 /M At m th remote the unwanted (CCI) signal arrives with power N(M 1)/M Capacity C = log 2 (1 + SINR) Hence.. ( C = log 2 1 + N ) M 1 (5) (6) 14 Coordinated and Distributed MIMO
Range I: M N: Interference limited: No ISI: Friis Transmission Assume CDM network scale D obeys D λ and the m th remote is located at R m D. Friisreceivedpower = G TG R λ 2 (4πR m ) 2 (7) Let G T = 1 and G R = 1. The range is distance at which P w = P u + P N. P w = N2 ( ) λ 2 ( ) N(M 1) λ 2, P u = (8) M 4πR m M 4πR m Range of CDM link ( λ 4πR m ) 2 = MP N N(N M + 1) (9) 15 Coordinated and Distributed MIMO
Range II: M N: Interference limited: No ISI: Friis Transmission Range of CDM link ( λ 4πR m ) 2 = MP N N(N M + 1) (10) Range of SISO link ( ) 2 λ = P N 4πR S (11) The ratio is... R m = R S N (N M + 1) (12) M 16 Coordinated and Distributed MIMO
Net Capacity: N = 100: ISI 50 taps 150 Net capacity (Bps/Hz) 100 50 0 0 50 100 150 200 Number of remotes 17 Coordinated and Distributed MIMO
Bit Error Rates: N = 100: ISI 50 taps 10 0 20 BER 10 2 10 4 SIR (db) 15 10 5 0 5 10 6 0 50 100 150 200 10 Number of remotes 18 Coordinated and Distributed MIMO
Technical and Protocol Issues Timing offset. Time reversal precoding requires that we coordinate the base stations on the downlink. CO-MIMO proposals have shown that IP network based timing sync wont work - GPS? Carrier offset. Easy? - GPS based or algorithm based with low phase noise local oscillators - Coherence time 1/(Phase noise bandwidth) 1sec - Easy? Time reversal precoding requires the channel state information be available to the bases. Assume reciprocity on the uplink to estimate the channel for the downlink - reciprocal transceivers? Use matched channel symbol decoding on the up-link - Identical to the downlink but all done in digital rather than the physical domain. Information transport - flap IP transport of signal samples on the backhaul link. 19 Coordinated and Distributed MIMO
Reciprocity 20 Coordinated and Distributed MIMO
Cellular Network Comparison I Consider a SISO cellular network with N cells, one base per cell and L-remotes per cell. B Hz of spectrum per remote Each remote requires a dedicated power P o from the base. To provide the same service we consider a CDM nerwork with NL base stations of EIRP P o in the first cell. 21 Coordinated and Distributed MIMO
Cellular Network Comparison II 22 Coordinated and Distributed MIMO
Cellular Network Comparison III: Results Cellular Distributed Number of BTs N LN User BW B B Total Occupied BW LB B Total BT P rad > LNP o LNP o P rad per BT > LP o P o Uplink BW per RT B B Range (99%) 2.15r o Lro 23 Coordinated and Distributed MIMO
Repeater Network 24 Coordinated and Distributed MIMO
AD-HOC 25 Coordinated and Distributed MIMO