Space Time Processing for Fixed Broadband Wireless A. Paulraj Gigabit Wireless & Stanford University ISART 6 -8 September, 2000 Boulder, CO

Essential Attributes of Broad Band Wireless (BWA) Network High Speed > 4-5 Mbps High QoS / Availability DSL / Cable like Low Costs $$$ High capacity High coverage CPE and Infrastructure unit costs Friendly Easy CPE Install Environmental/Regulatory Scalability (Multi-cell) Evolution to portability

BWA Network Architecture Multi-cell architecture Low BTS antennas ( ft) Under-the-eave subscriber antennas Significant Non-LOS propagation (Rayleigh fading) Strong multi-path Low Doppler channels Mobile cellular-like network with all its challenges but with  50 Speed and  50 QoS

NAU= Network Access Unit BTS = Base Transceiver Station BSC = Base Station Controller BWA Network NAU IP NGU BSC BTSBTS BTSBTS BTSBTS Voice gateway Network Operations Center SMS Access Mux PSTN IP Gateway NAU NGU= Network Gateway Unit SMS= Subscription Management System

Data Rate & QoS Challenges F  50 High Data Rate F  50 QoS (0.999 Availability on Throughput, Delay, Jitter) Capacity (Bits/Hz/Cell) Coverage (Cell Radius)  0.4  0.2  0.4

Typical Propagation Scenario BTS Directional Antenna BTS Ht 30’-150’ Ht 8’-20’ miles CPE Directional Antenna Distance to mobile scatterers

Path Loss (Mean Level) Signal Power (dB) Super Cell BTS ht: 1500ft Macro Cell BTS ht 150 ft Range in km Frequency: 2.5 GHz Ant Beamwidth: 90 deg Super Cell: Free Space model Macro/Micro Cell: Erceg’s model

K-Factor vs. Distance (Suburban BWA) ht = 15m, 90 deg. Rx antenna hr = 3m hr = 10m GW and AT&T measured data similar K = 0 is necessary assumption for reliable deployment --- hr = 10m --- hr = 3m

RMS Delay Spread vs Distance (Suburban BWA) RMS Delay Spread in Microseconds (dB) Distance in km GW experimental data similar

Doppler Spectrum - BWA - 3dB f max  0.05  f d  0.5 Hz

Signal Measurements Low Doppler, 0.05 Hz High Doppler, 0.5 Hz

C vs I Measurements

MIMO Measurements BTSCPE H22 H12 H21 H11

Spatial Multiplexing (SM) Offers Higher Speeds Each Eigen mode of the matrix channel can support a radio link  1.5 speed up possible in 2 x 3 systems. Tx Rx H

Increasing Capacity/Coverage Coverage : Lower the C/N needed to close link - improves link budget Capacity: Lower C/I needed to close link - improves reuse factor –Use diversity, array gain, coding, etc to minimize required C/N or C/I to support a target PER –Use RLP and fragmentation to allow higher target PER –Use spatial multiplexing to run parallel channels

Mode 6, Doppler 1Hz, MaxUtil 635 users, small files link utilization [%] mean throughput [Kbps ] Mean Per 1% Mean Per 5% Mean Per 10% Mode 6, Doppler 1Hz, MaxUtil 635 users, big files link utilization [%] mean throughput [Kbps] Mean Per 1% Mean Per 5% Mean Per 10% - TCP windowing algorithm yields the difference in the throughput of ‘small’ and ‘big’ files - Throughput = File size/ Mean transfer delay Throughput vs Utiliz. (vs PER)

Increasing Coverage Diversity is critical (Tx antenna, Rx antenna, frequency, pol) Directional CPE antennas Forward Error Correction (FEC) & ARQ Increase BTS & CPE height

Coverage vs Required C/N Range in Miles 1 x 1, NO RLP2 x 2, RLP Required C/ N

CDF of Tx-Rx Diversity - Measured Data Signal Power About Mean in dB Probability That Signal Power < Abscissa Before Combining After Combining 1 x 2 2 x 2 Probability of signal falling 10 dB below mean is 300 times lower with 2 x 2 compared 1 x 1

Increasing Spectrum Efficiency Reduce CCI to improve reuse - Increase signal diversity - Interference averaging - Antenna sectorization, beam forming, interference cancellation Reduce ACI to improve reuse - Careful frequency planning - Stringent emission masks, Rx filtering - Interference suppression Use link adaptation to exploit available C/N or C/I levels Use Radio Link Protocol to allow link to use high BER

C/ I After Combining - Measured Data C /I = 10 dB C/I About Mean in dB Probability That C/I < Abscissa Before Combining 1 x 2 2 x 2 After Combining Probability of C/I falling 15 dB below mean is 300 times lower with 2 x 2 compared to 1 x 1

Conclusion Achieving high capacity/coverage in a multi-cell BWA (  50 speed &  50 QoS compared to mobile cellular) network is a challenging problem. Many technologies help improve performance. MIMO antennas are a key leverage. Gigabit Wireless is a pioneer of MIMO for BWA networks