MIMO Wireless Systems Using Antenna Pattern Diversity Liang Dong Department of Electrical and Computer Engineering The University of Texas at Austin

April 5, /26 Outline Introduction and Motivation MIMO Wireless Communication The Innovation of Polarization Diversity MIMO & Antenna Pattern Diversity Conclusion and Ongoing Research

April 5, /26 Motivation Desire: –High bit rate wireless access. Problem: –Wireless throughput limited by scarce & expensive spectrum, power limitations, fading, interference, noise Goal: –Design spectrum efficient wireless links. –Spectrum efficiency = data rate / BW Solution: –Multi-antenna wireless communication systems.

April 5, /26 Multi-Antenna Wireless Systems Existing wireless systems (SISO / SIMO / MISO) –Multiple antennas at transmitter or receiver Future wireless systems (MIMO) –Multiple antennas at both transmitter and receiver

April 5, /26 Why MIMO Communication? Spectrum efficiency –Spectrum is expensive –Maximize data rate / bandwidth -> bits / s / Hz –MIMO technology can create multiple data pipes via spatial multiplexing Quality –Wireless links fluctuate due to fading and interference –Require high mean and low variance SINR for wire-like quality –MIMO technology can enable very reliable communication links Limited transmit power –Transmit power is limited in wireless systems –MIMO technology can reduce required transmit power and/or increase range/coverage

April 5, /26 What is a MIMO Communication Link? Propagation channel : channel matrix (narrow band) –For this talk: assume n T = n R = n. nTnT nRnR

April 5, /26 Channel Capacity with MIMO Channel capacity in bits per second per hertz: Mutual information (instantaneous channel capacity): MIMO capacity scales linearly as the number of antennas.

April 5, /26 Mutual Information in Real MIMO Channels Enormous (linear-scale) capacity exists when the channel matrix H unitary. Real channels are rarely unitary –Function of scattering environment –Antenna geometry What is the mutual information in practical channels? Decouple the MIMO channel into n SISO sub-channels: Performing SVD on H, let H = U  V. where, y = U + v (R), x = AVv (T), and u = U + n.

April 5, /26 Dependence Between Sub-channels The mutual information of the MIMO channel is the sum of the mutual information of the n SISO sub-channels: where,  i 2 is the gain of the i th sub-channel. Insufficient spacing results in loss of orthogonality between sub-channels. Any other ways to introduce more independent sub- channels?

April 5, /26 Polarization in MIMO Systems Dual-polarization gives 2 (mostly) orthogonal channels Cross-pole discrimination characterizes coupling –Typically 15dB in practical applications –Depends on terrain and environment Polarization diversity used in fixed wireless Antennas can be closely collocated Can we do better than 2x? LOS microwave links

April 5, /26 Mirror TransmitterReceiver Ex Ey Ez LOS Two degrees of electric field freedom LOS + Reflection A third degree of freedom results from the mirror (scattering). Multi-polarization Communication Links

April 5, /26 Communication in Multi-Polarization Channels Proposed by [Andrews et. al, 2001] H, 6 £ 6 channel transfer matrix. relating the electric (E) and magnetic (B) fields with idealized oscillating electric (p) and magnetic (m) dipole moments: = H

April 5, /26 Multi-Polarization Capacity Limits Define the number of polarization channels as rank(H). At large SNR, M(H) tends to the value rank(H) log2 (  ). Claim [Andrews et. al] In free-space, dual-polarized systems rank(H) = 2. In a scattering environment rank(H) = 6. => Tripling the capacity. Questions: Can we really get this improvement in real channels? Answer: This is a merely a loose upper bound in a real EM world.

April 5, /26 TX: (9.9, 7.7, 10.5), RX: (15.1, 109.8, 8.1), = m, same as in [Andrews et.]. TX RX Simulation (2-mirror environment)

April 5, /26 Antenna Radiation Patterns Infinitesimal electric-dipole (z)Infinitesimal magnetic-dipole (z) (current-loop) x y z xy z     E H E H

April 5, /26 Simulation Result (2-mirror environment) Eigenvalues of HH +. RX is at variable distance from the TX.

April 5, /26 MIMO Systems Using Pattern Diversity Consider a narrow band MIMO system, with closely collocated antennas at transmitter and receiver. Antenna pattern diversity appears in the transfer matrix.

April 5, /26 Building blocks (10  10  10 m) T 1 and T 2 : Transmission points. R 1 and R 2 : Receiving tracks. FASANT Simulations

April 5, /26 Case 1: T 1 => R 1 (both LOS and NLOS regions) Eigenvalues of HH +

April 5, /26 Case 1: Comparison of Mutual Information Compare (local-averaged) M(H) of 6 £ 6, 3 £ 3 and 2 £ 2 MIMO channels. The LOS region is y 2 [13.33, 26,67] m.

April 5, /26 Case 1: Comparison of Mutual Information Ratios of M(H) of 6 £ 6, 3 £ 3 and 2 £ 2 MIMO channels.

April 5, /26 Case 2: T 2 => R 1 and R 2 (Separated LOS and NLOS cases) Receiver on LOS street Receiver on NLOS street

April 5, /26 Case 2: Caparison of Mutual Information Receiver along: (a) R 2 (LOS street ). (b) R 1 (NLOS street).

April 5, /26 Case 2: Caparison of CCDFs Receiver along: (a) R 2 (LOS street ). (b) R 1 (NLOS street).

April 5, /26 Conclusions MIMO systems that exploit antenna pattern diversity allows improvement for closely collocated receiver and transmitter antennas. The capacity increase is limited by the sub-channel correlation in a real electromagnetic world. The capacity increase depends on the characteristics of the scattering environment.

April 5, /26 Ongoing Research Design of compact antennas: Antenna radiation pattern selection for optimal MIMO performance. Propagation channel study. Analysis of channel correlation and channel capacity of MIMO system that exploits antenna pattern diversity. UT Faculty Contacts –Prof. Hao Ling – antenna design & analysis –Prof. Robert Heath – MIMO communications