Achieving High Data Rates in a Distributed MIMO System Horia Vlad Balan Ryan Rogalin Antonios Michaloliakos Konstantinos Psounis Giuseppe Caire USC
Structure of this talk Motivation Multiuser MIMO and precoding schemes Distributed MIMO and synchronization Experimental results
[Webb - The Future of Wireless Communication] Motivation Cellular companies spend billions for more bandwidth Spectrum reuse is the most promising way to increase wireless transfer rates and distributed MIMO is its ideal implementation In WiFi networks, with a high number of users, spectrum reuse becomes equally important [Webb - The Future of Wireless Communication]
Enterprise WiFi
Multiuser MIMO
Shannon’s Theory
Increasing the Rate Prelog Factor Increase your bandwidth! Inlog Factor Increase your power exponentially!!!
MIMO Communication interference
Dirty Paper Coding provides the achievable rate region Separate the Channels limited interference Dirty Paper Coding provides the achievable rate region
Zero-Forcing -1
Tomlinson-Harashima Precoding U -1 L U L
Tomlinson-Harashima 3 3 -1 1 2 +2 -2 1 +2 +3 +4 -9 (mod 5) = 1 4 Modulo Compensation -3 1 2 3 4 5 6 7 8 9 -2 -1 -4 -5
Tomlinson-Harashima Precoding U -1 ) mod ( mod L U -1
Blind Interference Alignment + 3 slots, 4 symbols => 4/3 DoFs + 72 + + -1 + +
Distributed MIMO
Challenges Maintaining phase synchronization between the different APs Gathering channel state information and transmitting before the channel coherence time ends
OFDM Modulation Subcarriers Carrier Cyclic Prefix OFDM Symbol
OFDM Demodulation IFFT FFT
Distributed OFDM Symbol Alignment TX 1 Phase Alignment TX 2 RX FFT
Distributed OFDM TX 1 TX 2 Carrier Frequency Random Timing Offset Phase TX 1 TX 2
Phase Alignment
Phase Alignment What should be the effective channel matrix? option 1 option 2: coherence time depends on the electronics What should be the effective channel matrix?
What should be the effective channel matrix? Phase Alignment option 1 What should be the effective channel matrix?
Achieving Phase Synchronization Pilot Signal Data User Master Secondaries
Distributed MIMO Testbed Pilot Signal Master TDMA point-to-point Secondaries Data Clients (4x4 MIMO)
Channel Orthogonalization Results Phase Accuracy Channel Orthogonalization ZFBF (2x2 MIMO)
(85% of the theoretical gain) Results Tomlinson Harashima 85% rate increase (85% of the theoretical gain) (2x2 MIMO)
(55% of the theoretical gain) Results Tomlinson Harashima 165% rate increase (55% of the theoretical gain) (4x4 MIMO)
Results Blind Interference Alignment 22% rate increase (66% of the theoretical gain)
MAC Layer Results Comparing scheduling strategies through simulation in a 4 AP, 8 users scenario Greedy Zero-Forcing, Tomlinson-Harashima precoding, Blind Interference Alignment Using TDMA as a reference point .
4x4 achievable rates (simulation) Results 4x4 achievable rates (simulation)
Future Work improving the accuracy of our estimators combining distributed MIMO with incremental redundancy schemes characterize the channel quality variations of BIA in large deployments
Questions?
Thank you!