Adaptive Antenna Tutorial: Spectral Efficiency and Spatial Processing

Adaptive Antenna Tutorial: Spectral Efficiency and Spatial Processing Project MESA Meeting 30 October – 2 November 2006 Joanne Wilson VP, Standards ArrayComm, LLC.

ArrayComm: Industry Leadership • ArrayComm background • World leader in adaptive antenna technology (also referred to as “Adaptive –Multi-Antenna Signal processing” (A-MAS) or “smart antenna” technology) • Founded 1992 • Over 300,000 base stations deployed • Extensive patent portfolio in A-MAS Technology • $250M invested in technology development & commercialization • Technology and end-to-end systems • IntelliCell A-MAS technology for PHS, GSM, W-CDMA, 802.16 • End-to-end wireless systems including HC-SDMA, WLL • Consistently reducing costs of coverage and capacity • Business model • Software products, services • Technology development, transfer • Chipsets

The spectral efficiency bottleneck • Today’s principal spectral inefficiency • omnidirectional radiation and reception • Why? • tiny fraction of power used for communication • the rest: interference for co-channel users • So: • Exploit spatial properties of RF signals • Provide gain and interference mitigation • Improve capacity/quality tradeoff • And… What is spectral efficiency? • bits/seconds/Hz/cell • Measures how well a wireless network utilizes radio spectrum • Determines the total throughput each base station (cell) can support in a network in a given amount of spectrum New air interfaces should be built from the ground up to be optimized for spatial processing

802.11b A-MAS Benefit 2/2.5/3G The Capacity/Coverage Tradeoff Throughput/cell (Mbps) Technical Interpretation • Gain vs. noise, fading, ... expands envelope to right • Interference mitigation (+ gain) expands it upwards Economic Interpretation • Coverage improvements reduce CapEx, OpEx (esp. backhaul, sites) • Capacity improvements reduce delivery cost, spectrum requirements range (km) Interference Limited Noise Limited

Motivation • Wireless system design is a trade-off of competing requirements • service definition • service quality • capacity • capital and operating costs • resource requirements including spectrum • end-user pricing/affordability • coexistence with other radio technologies • A-MAS technology fundamentally changes the nature of this trade-off and achievable system performance

802.11n 802.11 WiFi W-LAN 802.16 MMDS/FWA Cable/DSL MBWA ANSI/ATIS HC-SDMA (iBurst) 802.20 3GPP, 3GPP2 802.16d 3G 3G 802.16e LTE CL/OL Diversity, MIMO Adaptive Antennas in All New Broadband Systems ITU Recommendation M.1678 (2004): “This Recommendation considers the ability of adaptive antenna systems to improve the spectral efficiency of land mobile networks and recommends their use in the deployment of new and the further enhancement of existing mobile networks. It also recommends the integration of this new technology into the development of new radio interfaces.” User Data Rate Satellite Dialup Fixed Local Area Wide Area Mobility

Outline • Spectral Efficiency and System Economics • Adaptive Antenna Basics • Adaptive Antenna Technologies • Adaptive Antenna Performance Determinants • Summary • Backup Slides

Spectral Efficiency Defined • A measure of the amount of information – billable services – that carried by a wireless system per unit of spectrum • Measured in bits/second/Hertz/cell, includes effects of • multiple access method • modulation methods • channel organization • resource reuse (code, timeslot, carrier, …) • “Per-Cell” is critical • fundamental spectral efficiency limitation in most systems is self-generated interference • results for isolated base stations are not representative of real-world performance

Why Is Spectral Efficiency Important? • Spectral efficiency directly affects an operator’s cost structure • For a given service and grade of service, it determines • required amount of spectrum (CapEx) • required number of base stations (CapEx, OpEx) • required number of sites and associated site maintenance (OpEx) • and, ultimately, consumer pricing and affordability • Quick calculation offered load (bits/s/km2) number of cells/km2 = available spectrum (Hz) x spectral efficiency (bits/s/Hz/cell)

Increased Spectral Efficiency • Increased spectral efficiency leads to • improved operator costs • reduced equipment CapEx/OpEx per subscriber • reduced numbers of sites in capacity limited areas • reduced barriers to new operators • better use of available spectrum • especially important for limited mobility spectrum • improved end-user affordability, especially for broadband services • Spectral efficiency will become even more important • as subscriber penetration increases • as per-user data rates increase • as quality of service (esp. data) requirements increase

Spectral Efficiency Design Elements • Spectral/Temporal elements • multiple access method: TDMA, FDMA, OFDMA… • optimize efficiency based on traffic types • modulation, channel coding, equalization: QPSK, QAM, OFDM, … • optimize efficiency based on link quality • Spatial elements (all to minimize interference) • cellularization • mitigate co-channel interference by separating co-channel users • sectorization • mitigate co-channel interference by more selective downlink patterns and increased uplink sensitivity • power control • use minimum power necessary for successful communications

Increasing Spectral Efficiency • Temporal/Spectral issues are mature, well understood, well exploited • no significant future improvements in spectral efficiency here • proper application is important • Least spectrally efficient aspect of most systems • omnidirectional/sectorized distribution and collection of radio energy • Why? • Most of the energy is wasted. • Worse, it creates interference in the system and limits reuse.

Sectorized Transmission/Reception interference • Sectorized, spatially non-selective, transmission causes interference in adjacent cells • Similarly, increases sensitivity to interference from adjacent cells • Cellular “reuse” mitigates this effect by separating co-channel users • Cost: decreased resources per sector and reduced spectral efficiency cells sectors serving sector user

How Do Adaptive Antennas Help? • Adaptive antennas are spatial processing systems • Combination of • antenna arrays • sophisticated signal processing • Adapt the effective pattern to the radio environment • users • interferers • scattering/multipath • Provide spatially selective transmit and receive patterns

Adaptive Transmission/Reception • Spatially selective transmission reduces required power for communication • Decreases sensitivity to interference from adjacent cells • Allows reuse distances to be decreased • possible to reuse resources within a cell in some cases • Benefits: increased resources per sector, increased spectral efficiency cells sectors serving sector interference user

Adaptive Antennas Defined • Systems comprising • multiple antenna elements (antenna arrays) • coherent processing • signal processing strategies (algorithms) that vary the way in which those elements are used as a function of operational scenario • Providing • gain and interference mitigation • leading to improved signal quality and spectral efficiency

Adaptive Antenna Fundamentals • Solution elements • multiple antenna elements and transceiver chains • scenario-dependent signal processing • air interface support for highest performance, e.g., training • Link-level performance benefits • diversity • gain • interference mitigation

SISO, MISO, SIMO, MIMO, … SISO • Single Input, Single Output MISO • Multiple Input, Single Output SIMO • Single Input, Multiple Output MIMO • Multiple Input, Multiple Output SDMA

Adaptive Antenna Gains (transmit or receive) Diversity • differently fading paths • fading margin reduction • no gain when noise-limited Coherent Gain • energy focusing • improved link budget • reduced radiation Interference Mitigation • energy reduction • enhanced capacity • improved link budget Enhanced Rate/Throughput • co-channel streams • increased capacity • increased data rate

8x: 12 dB reduction 2x: 7 dB reduction Diversity • Slope of error curve proportional to diversity order (# antennas) • Transmit/receive channel knowledge not required • Reduces required fading margin • Selection diversity • Single Tx antenna • Independent fading 1 antenna 8 antennas

Going Further: Gain, Capacity, QoS, Data Rate • (Multi)Channel state information (CSI) required to go further • coherent gain, interference mitigation, capacity/rate increases • Theoretical SNR gain with M antennas: M or 10log10M dB • achievable in practice with good design, esp. for receive processing • Rx and Tx • Theoretical interference rejection is infinite • limited in practice by scenario, protocol, equipment. • 20 dB for significant interferers readily achievable • New protocols include training/feedback for spatial processing • analogous to training for equalizers

Adaptive Antenna Concept • Users’ signals arrive with different relative phases and amplitudes at array • Processing provides gain and interference mitigation User 2, s2(t)ejt User 1, s1(t)ejt as1(t)+bs2(t) as1(t)-bs2(t) +1 -1 +1 +1 2as1(t) 2bs2(t)

Protocol Independence • Fundamental concepts applicable to all access methods and modulation methods … antenna antenna … Transceiver Transceiver … Channelizer (TDMA, FDMA, CDMA) Channelizer (TDMA, FDMA, CDMA) Spatial and Temporal Processing baseband signals/user data

Interference Mitigation • Gain and interference mitigation performance are actually statistical quantities • Theoretical gain performance closely approached (within 1 dB) in practice • Theoretical interference mitigation, , harder to achieve • limited by calibration, environment, number of interferers • Practically, active mitigation in excess of 20 dB can be achieved for significant interferers • Active interference mitigation independent of and in addition to gain • Directive gain term generally results in some passive interference mitigation

Comments • Fundamental concept is coherent processing • Generally applicable to all air interfaces • Processing is done in parallel on all traffic resources • Line-of-sight is not required • Many important issues that can’t be addressed here • estimation of radio environment (algorithms) • processing requirements (easily > 1Gbps of data from the array) • performance validation • equipment calibration • effects of air interface specifics (will comment on this later) • reliability benefits of redundant radio chains • intrinsic diversity of an array

Antenna Arrays • Wide variety of geometries and element types possible • arrangements of off-the-shelf single elements • custom arrays • Array size • vertical extent determined by element gain/pattern as usual • horizontal extent, typically 3-5 lambda • Array of eight 10 dBi elements at 2 GHz is about 0.5 x 0.75 m • small! • conformal arrays for aesthetics

Processing At The User Terminal • This presentation focuses on adaptive antennas at the base station • Adaptive antennas can also be incorporated at the user terminal • base station and user terminal can perform independent adaptive antenna processing • base station and user terminal can perform joint adaptive antenna processing, so called “MIMO” systems, with additional benefits • Fundamental issue is an economic one • incremental costs at base station are amortized over many subscribers • incremental costs at user terminal are amortized over one user, solutions must be inexpensive for consumer electronics applications

Adaptive Antenna Potential Processing Gain Operational Significance Selective Uplink Gain Increased Range & Coverage Increased Data Rates Reduced System – Wide Uplink Noise Improved Uplink Multipath Immunity Improved Signal Quality Maintained Quality with Tightened Reuse Increased Range & Coverage Increased Data Rates Reduced System–Wide Downlink Interference Improved Co–existence Behavior Reduced Downlink Multipath Maintained Quality with Tightened Reuse Uplink Interference Mitigation Selective Downlink Gain Downlink Interference Mitigation

Adaptive Antenna Technologies (1) • Actual level of benefits depends on details of the implementation, little variation in general hardware architecture across implementations • Basis for comparison • predictability and consistency of performance • balance of uplink and downlink performance (key for capacity improvements) • downlink is generally most challenging aspect of adaptive antennas • base station directly samples environment on uplink; must infer the environment on the downlink • robustness of performance across variations in propagation and interference scenarios

Adaptive Antenna Technologies (2) • Switched Beam • selects from one of several patterns based on power • can be thought of as micro-sectorization • predictable gain and scenario-dependent interference mitigation in positive C/I environments • peak gain typically traded off for in-sector gain uniformity • variant: cell sculpting, select from several patterns for load balancing • Adaptive Energy Extraction • attempts to extract maximum energy from radio channel • maximal ratio and combined diversity are examples • scenario-dependent gain and interference mitigation in positive C/I environments • gain near theoretical maximum in high SINR environments

Adaptive Antenna Technologies (3) • Model-Based or fully adaptive • continuous adaptation based on model including users and interferers • simultaneous gain and active interference rejection possible, even at low SINR’s • manageable increase in computation as compared to other methods • availability of channel assignments and other high-level protocol information improve performance

Adaptive Antenna Performance • Primary determinants • environmental complexity • degree of mobility • duplexing: frequency-division or time-division (FDD vs. TDD) • issue is correlation of uplink and downlink propagation environments • Capacity increases in operational systems

Comparing TDD and FDD • Advantages and disadvantages to both

Summary • Increased spectral efficiency leads to • better spectrum conservation • diversity of services • affordability of services • A-MAS is the single best technology for increasing spectral efficiency • Wide range of A-MAS technologies • same basic principles • wide variations in goals and performances • intracell reuse (reuse < 1) possible for certain applications • Proven technology • more than 300,000 deployments worldwide

End-User Affordability • Example • A wireless operator charges $60/mo. for 450 minutes of 10 kbps speech over system A, about $0.22/Mbit • Another wireless operator charges about $500/mo. for 1 Gbyte/yr over system B, about $0.75/Mbit • similar spectral efficiency for systems A and B, similar operating costs, similar price/bit • advanced, high-speed, services are not affordable for most end-users at this spectral efficiency • Important point, although oversimplified example • data and voice network and service costs differ • new equipment cost must be recaptured • 1 Gbyte/yr is casual primary internet access, operators may be trying to discourage this use of their network

Basic Uplink Gain Calculation • Signal s, M antennas, M receivers with i.i.d. noise ni • Adaptive antennas provide uplink gain of M or 10log10M dB • M=10, 10x SNR improvement, examples • double data rate if single antenna SNR is 10 dB • reduce required subscriber transmit power by 10 dB • increase range by 93% with R3.5 loss received signal s + ... + s = noise n1 + … + nM (Ms)2 s2 therefore, Uplink SNR = = M Ms2 s2 = M x single antenna SNR

(P/M s + … + P/M s)2 ( Ps)2 Basic Downlink Gain Calculation • Similar to uplink calculation, except dominant noise is due to (single) receiver at user terminal • With same total radiated power P in both cases • Again, factor of M or 10log10M dB • M=10, 10 dB gain examples • 10 element array with 1 W PA’s, has same EIRP as single element with 100 W PA • For given EIRP can reduce total radiated power by 10 dB, 90% interference reduction Received Power (AA) = = M Received Power (SA)

Spatial processing creates unique advantage System Capacity System Range Mobile Wireless System Capacity in Mature Networks, Mbps aggregate BTS capacity per MHz available HC-SDMA 4.0 802.16+A-MAS* 1.7 With A-MAS (i.e smart antennas) MC-SCDMA(proprietary variant) 1.2 PHS+A-MAS* 0.7 GSM+A-MAS* 0.6 WCDMA+A-MAS* 0.4 EV-DO or HSDPA 0.4 Flash-OFDM 0.4 802.16 0.3 TD-CDMA 0.2 Without WCDMA 0.2 GSM 0.1 PHS 0.04 *Standard protocol with base station enhanced by A-MAS technology Sources: Vendor claims for maximum BTS throughput, ArrayComm field experience in Korea and Australia, various analysts.

Some Comparisons 40 Cell Capacity Throughput in 10 MHz (Mbps) System Spectral Efficiency 4.0 3.5 3.4 2.2 2.1 3.0 0.16 2.5 GPRS CDMA2000 WCDMA 1xEV-DO HC-SDMA 2.0 Spectral Efficiency in bits/sec/Hz/cell 1.5 250 Network Capacity Number of cells to deliver the same information density, Mbps per KM2 1.0 0.5 0 PHS WLL GSM GSM IS-95B IS-95C IS-95 A HSCSD 19 18 HC-SDMA IntelliCell® IS-95 HDR Cdma2000 12 1 GPRS CDMA2000 WCDMA 1xEV-DO HC-SDMA

Adaptive Antenna Performance Adaptive antennas benefit all systems, but HC-SDMA extracts maximum benefits by design

Co-Channel Regulatory Issues • Recall adaptive antennas’ high ratio of EIRP to total radiated power (TRP) • factor of M higher than comparable conventional system • result of directivity of adaptive antennas • Average power radiated in any direction is then TRP plus gain of individual array elements (worst case directive power remains EIRP) • Relevant in setting EIRP limits for coordination of co-channel systems in different markets • Very relevant in RF exposure considerations

Adjacent Channel/Out-Of-Band Regulatory Issues • Recall that adaptive antenna gains result from coherent processing • Out-of-band radiation due to intermodulation, phase noise, spurs • nonlinear processes • reduce/eliminate coherency of signals among PAs’ out-of-bands • Result • ratio of in-band EIRP to out-of-band radiated power is up to a factor of M less than for comparable conventional system • Rules may want to anticipate adaptive antennas • A per-PA “43+10logP-10logM rule” would result in comparable operational out-of-bands as single antenna 43+10logP rule • significant positive effect on adaptive antenna power amplifier economics • may help to foster adoption

iBurst (HC-SDMA) Highlights • Time division duplex (TDD) • Packet switched TDMA/SDMA multiple access scheme • Adaptive modulation & coding • Fast ARQ for reliability, low latency • Peak per-user rate 16 Mbps (initial products support 1 Mbps peak) • 40 Mbps throughput in 10 MHz (DSLAM equivalent) • Centralized resource allocation for efficiency, QoS • Inter-cell and inter-system (e.g., 802.11) handover • Standardized by American National Standards Institute (ANSI) • ANSI ATIS 0700004-2005, High Capacity-Spatial Division Multiple Access (HC-SDMA) • Soon to be officially recommended by the ITU-R • Included in Draft New Recommendation ITU-R M.[8A-BWA]

iBurst Frame and Traffic Bursts • iBurst uplink/downlink traffic slots paired • spatial+temporal training

Cross Layer Design: Spatial Processing MAC Multiple logical channels per physical resource • paging and/or traffic and/or access • Spatial collision resolution • enables low latency/low jitter designs Traffic Traffic UT UT UT Page BS BS UT Traffic Access UT

Adaptive Antenna Tutorial: Spectral Efficiency and Spatial Processing