Ultra-Widebad (UWB) Wireless Commuicatios Associate Professor Teessee Techological Uiversity rqiu@ieee.org Preseted at Army Research Lab Adelphi, Marylad, Jue 4, 2004 1
Outlie Itroductio UWB Priciples IEEE 802.15.3a/IEEE 802.15.4a Receiver Desig Challeges OFDM ad Pulse-based UWB Physics-Based Optimum Receiver Structures Commuicatio Theory ad Physics (Gabor( 1953) Coclusio 2
Established Aug. 1 2003 UWB/3G/4G physical layer ad cellular etwork levels Radio Propagatio ad Chael Modelig Receiver aalysis ad desig Hardware prototypig Workig with idustrial ad DOD R&D orgaizatios. 10+ years R&D experieces i wireless commuicatios 3
Mobile Devices Market Segmetatio odules - mbedded Apps elematics / elemetry dd-o evices ata Devices / Itegral ireless usiess / mart hoes asic hoes Nokia 3330 PDQ Smart Phoe Ericsso R380 RIM Blackberry HP Jorada 720 w/ PC Card Palm HadSprig Visor, Sprig Board Modules Greater Multi-Media Capability Larger Displays / Touch-Screes ad Keyboards Multi Wireless Modes & Geerally Higher Data Rates 4
3G & UWB Combiig Air Iterface Complemetig Techologies 3G Local Area Network Persoal Area Network WLAN UWB WCDMA EDGE CDMA2000 Wide Area Network Not to Scale 1 Wide Area cell = ~10 000 WLAN cells 5
The UWB Home Wireless Network Broadbad services: Cable, xdsl,satelite, Terestrial Home Gatew ay IEEE 802.15.3a Moitor Desktop computer Camcorder Priter DVD TV Laptop computer Digital Camera PDA Audio 6
UWB Commuicatios & Sesor Networks Eviromets Real-time Distributed Dyamic Hostile Applicatios Remote surveillace, threat detectio Video to the foxhole/battlefield High-resolutio locatio services Key Techologies Ultra-wide bad systems Mobile, adhoc etworks Data fusio / sythesis Ope Research Issues Pulse Propagatio Optimum Receiver Test-bed developmet / trials IEEE 802.15.4a DARPA Networkig i Extreme 7 Eviromets (NETEX)
What Is Ultra-Widebad (UWB)? Defiitio (I radar,etc) f u -f l Or greater tha 500 MHz (FCC Feb 2002) At FCC Part 15 powers (a few tes of microwatts total - across several GHz), caot be reliably measured below 10 db dow poits 2 0.25=25% f u +f l Where: f u = upper 10 db dow poit f l = lower 10 db dow poit 8
.5 GHz UWB Spectrum Allocated by FCC 02/200 9
Time Modulated Ultra-Widebad Widebad A Not a siewave,, but millios of pulses per secod Example 500 ps Time coded to make oise-like Chaelizatio Ati-jam Smooths spectrum Amplitude Radomized Time Codig Time 0 1 Power Spectral Desity (db) 0-40 -80 Frequecy (GHz) Radom oise sigal 1 2 3 4 5 Frequecy (GHz) Pulse positio modulatio δ δ δ = 125 ps 10
UWB FCC allocated 7.5 GHz ulicesed spectrum (2002) Requires shift i thikig Short Pulses are buildig blocks. Fadig is ot a major issue Too may resolvable quasi-static static pulses Pulse distortio Determiistic solutios from Maxwell s equatios UWB radio may be good for low data rate (<a few Mb/s) applicatios (IEEE 802.15.4a) 11
Experimetal Setup 0.5 s pulse 12
UWB Pulse Spectrum 13
Idoor 14
Flat grass grou Reflectio from buildigs Outdoor 15
Represetative Measuremets (USC) Office Rcvd Blocked LoS Hold Rcvd Clear LoS Hold Rcvd Blkd LoS 5 6 2.5 0 4 50 s 200 s 2 1.5 200 s 5 2 1 0 0.5 0 0-2 -0.5-5 -4-1 0-6 -1.5-2 5 1000 1050 1100 1150 1200 1250 time (aosecods) -8 2200 2400 2600 2800 3000 3200 3400 3600 3800 4000 time (aosecods) -2.5 2200 2400 2600 2800 3000 3200 3400 3600 3800 400 time (aosecods) 5 6 2 0 1 s 4 1.5 5 2 1 0.5 0 0 0-5 -2-0.5 0-4 -1 5 984 985 986 987 988 989 990 991 992 993 time (aosecods) -6 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 time (aosecods) -1.5 2070 2071 2072 2073 2074 2075 2076 2077 2078 20 time (aosecods) 16
Per-Path Path Pulse distortio Chael distortio Pulse delay per-path path pulse distortio A ew pheomeo for UWB. Caused by frequecy-selectivity of the chael medium. 17
UWB Receiver Desig Challeges Eergy collectio versus complexity (cost) RAKE may be too costly Time sychroizatio Iter-symbol iterferece (ISI) 10 symbols overlappig for idoor (100 Mbps) Symbol-level level equalizer No-coheret detector Trasmitted referece Eergy-detector 18
Why UWB ad why spectrum agility? Why UWB for IEEE 802.15.3a? UWB techology is uiquely suited for high-rate, short rage access» Theoretical advatages for approachig high rates by scalig badwidth» Newly allocated ulicesed spectrum (7.5 GHz) that does ot take away from other arrowbad systems (licesed or ulicesed)» CMOS implemetatios ow possible at these higher frequecies All CMOS architecture Why spectrum agility for a UWB solutio? Just because the FCC allows UWB to trasmit o top of other services does ot mea we should!» Govermet regulatios should be broader tha idustry requiremets Spectrum usage ad iterferece eviromet chages by coutry locatio, withi a local usage area, ad over time» Eable adaptive detectio ad avoidace strategies for better coexistece ad possible o-cotiguous spectrum allocatios for flexible regulatios i future Allow for simple backward compatibility ad future scalability 19
Flexible Spectrum Use Uexpected Iterferer Low Frequecy Set Group High Frequecy Set Group ~ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 3.1 10.6 Drop bad i Sacrifice sub-bad for coexistece Japa Drop bad iterferece Europe mitigatio (based o regulatio ad geographical locatio) Reserved Ceter frequecies chose for ease of implemetatio 440 MHz bad separatio for improved flexibility ~538 MHz wide bads to best utilize spectrum 20
Commuicatios ad Physics (Gabor 1953) The electromagetic sigals used i wireless commuicatio are subject to the geeral laws of radiatio ad propagatio. Commuicatio theory developed maily mathematical lies, takig for grated the physical sigificace of the quatities which figure i its formalism. Commuicatio is the trasmissio of physical effects. Hece commuicatio theory should be cosidered as a brach of physics. 21
Physics-Based Chael Model ad Optimum Receiver Structures Goal: Coect the time-domai electromagetics ad commuicatio (iformatio) theory. Missio: Develop the optimum detectio theory of physica sigals govered by Maxwell s Equatios. Tasks: Chael model models based o experimets ad theory (aalytical ad computer simulatios) Optimum detectio theory ad iformatio theory Sub-optimum receivers Hardware system (trasceiver) prototypig 22
er-path Pulse Distortio Based UWB Chael 1 c t 2 2 ( ) E( tt, '; rr')=-, δ( t-t') δ( r-r' ) 2 2 2 2 ( + k ) E( k, rr', ) = δ ( r-r' ) Rx A B A N α H( ω) = A ( jω) e h = 1 1 jωτ N α 1 ( τ ) = A τ δ( τ τ) = 1 Γ ( α ) α = 1 2 for a sigle edge diffractio Multiple diffractio must be icluded! 23
Cocept of UWB Pulse Distortio due to Diffractio N N N N 1GO 2 GO GTD GO / GTD h( τ ) = A δ( τ τ ) + B R ( τ) δ( τ τ ) + C g ( τ) δ( τ τ ) + D [ R ( τ) g ( τ)] δ( τ τ ) = 1 = 1 = 1 = 1 Wedge Wedge N α H( ω) = A ( jω) e = 1 jωτ Groud N 1 α 1 h( τ ) = A τ δ( τ τ) = 1 Γ ( α ) Distorted Pulse UWB pulse distortio is a physical pheomeo!!! 24
Physics-Based Multipath Chael Model δ ( τ ) τ τ 1 τ τ 2 τ τ N δ ( τ ) j 1 A e φ j 2 A e φ 1 τ τ 1 τ τ 2 j 1 A e φ j 2 A e φ 1 2 2 j N AN e φ τ τ j N AN e φ N h ( τ ) h ( τ ) ( ) 1 2 hn τ Turi s Model Sice 1956 N j h() τ = Ae φ δτ τ = 1 h ( τ ) = per path impulse respos Qiu 1995 ( N j h() τ = Ae φ h() τ δ τ τ 25 Robert = 1 Qiu (
Diffractio-Based Pulse Shape Distortio 1.5 Diffracted Sigal d(t)ad Template Sigal v(t) 1 0.5 d(t) ad v(t) 0-0.5 alpha= -1: 0.25: 0 (bottom to top) -1 -- alpha=0 <==> Icidet Waveform Red dashed Template Pulse v(t) -1.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 time t (s) 26
Per-Path Path Impulse Respose (Characterizig the Pulse Waveform) h ( τ ) C ξ ( τ τ) ( τ τ), τ τ = D η( τ τ) ( τ τ), τ > τ α α < α = 0! α α α = 0! D 1 t t C 1 t t H ( ω) = η( ) t e dt ξ( ) t e dt 0 0 0!( jω) = jω!( jω) jω Example: ξ ( τ) = 1/ τ η( τ) = 1/ τ Pulse diffracted by a PEC Edge 27
Compariso of Exact Solutio with Asymptotic GTD/UTD Solutios 1 0.8 0.6 Direct=0.35 Reflected = 0.65 Diffracted= 0.92735 UTD Keller Felse Exact 0.4 Cross-Correlatio 0.2 0-0.2-0.4-0.6-0.8-1 0 0.5 1 1.5 2 2.5 3 3.5 4 time t (s) 28
UWB Pulse Shape Trasform Caused by Diffractio 1 0.8 0.6 Alpha=0 Alpha=-1:0.25:0 0.4 Alpha=-1 Cross-Correlatio R xy (t) 0.2 0-0.2-0.4-0.6-0.8-1 -6-4 -2 0 2 4 6 time t α H ( jω) = ( jω) 29
Physics-Based Optimum Receiver Structures recived sigals rt () Matched Filter * y(-t) sampler MLSE t = T ( Viterbi) s data output rt () = ayt ( T) + t () = Iter-symbol Iterferece or Multiuser Detectio s yt () = xt () ht () N ht () = Ah() t δ ( t τ ) = 1 xt ( ) = trasmitted pulse shape 30
Summary UWB is oe of the most promisig techologies 7.5 GHz ulicesed spectrum from 3.1-10.6 10.6 GHz Volume products will be shipped i 3-43 4 years UWB is good for both short-rage rage (10-30m) ad log-rage (100-1000m) 1000m) Per path pulse distortio i a UWB chael is oe of the majo potetial problems i system desig Experimetal measuremets verified 31
Thak You! rqiu@ieee.org 32