1. The Hierarchy of Wireless Data Networking (An RF View) Dan Dobkin and Titus Wandinger WIC 2002

2. The Communications Edge TM Overview Performance of wireless links is limited by fundamental and practical constraints: –“beam” capacity BW*log(S/N) [Shannon] –Geographic reuse and interference management –Watts and the cost to deliver them Different technologies are optimal for different needs –UWB for short range ultrahigh rate –WLAN for moderate range high rate –WAN for long range low rate –SatCom for coverage Each technology has a different set of problems Ideal future network will combine where applicable

3. The Communications Edge TM Ultrawideband UWB is just spread-spectrum with an attitude –Transmit signals spread across several GHz of bandwidth Right on top of existing licensed and unlicensed users –Keep power low enough to avoid excessive interference to other users of the same spectrum

4. The Communications Edge TM Ultrawideband Performance Limits UWB range limited to about 2x(distance to nearest non-participating radio) for high datarates –Note that at low data rate you get more processing gain 100 Kbps = 45 dB => much longer range CONCLUDE: UWB is ideal for short distances and high datarates; low power = low cost (someday)

5. The Communications Edge TM Ultrawideband: RF Challenges Interference: –UWB “stomps on” users in many bands –FCC constrained operation to >3 GHz to avoid interference with GPS and air-traffic-control radar –Adaptive operation may enable higher power operation => greater range at high datarate Is this desirable? –Dithering of pulsed signals necessary to avoid narrow band radiation at multiples of (1/pulse rate) UWB self-interference –Code-division multiplexing is also versatile solution for high processing gain systems Decoding –Power needed to encode / decode the signal remains large even if RF power is very small Ex: Xtreme Networks chipset 200 mW for < 1 mW RF

6. The Communications Edge TM Wireless LAN WLAN = Wireless Ethernet –Data-only in unlicensed bands (2.4 or 5.x GHz); nominally spread- spectrum [but not really] –Range limited not by technology but (mostly) by regulatory limits on transmission power COMMERCIAL EQUIPMENT TODAY: –6-30 Mbps TCP thruput, 50 meter indoor range –VERY low cost due to simplicity, high volume, low power –Conclude: WLAN provides high data rates over moderate distances at low capital cost

7. The Communications Edge TM WLAN RF Challenges Partitioning / coverage: –RF propagation in buildings is complex –Roaming adaptation in 802.11 networks is limited –Installation should be simple and inexpensive –Constraints are not consistent! Interference: –Multiple users (esp. in a building) may interfere –802.11 (CSMA/CA) is robust to interference but rate drops –Smart antennas would help: Most transmissions from MS are directed to a particular AP Some transmissions from AP are directed to a particular MS In each case an adaptive antenna will deliver more signal and less interference with less power consumed

8. The Communications Edge TM WAN: “cellular” communications Voice network modified for data communications –EXAMPLE: Qualcomm’s CDMA2000-EVDO 1.25 MHz channel devoted to data only; forward link is time- domain multiplexed, reverse link is code domain multiplexed Forward link capacity 2.5 Mbps/basestn for slow-moving terminals; reverse link capacity about 0.5 Mbps Typical range 1 km at high rate (adaptive) –Sophisticated handoff, authentication, billing –High capital cost –Key problem is economic not technical: Low-revenue-per-bit data displaces high-revenue-per-bit voice Conclude: good coverage, low data rate, high capital cost

9. The Communications Edge TM WAN RF Challenges Power consumption: –Large range achieved by use of high transmit power –High data capacity achieved by use of advanced modulations Stringent requirements on linearity => serious backoff from maximum output power –Result: use 1 KW to transmit 100 W RF High hardware cost, high operating cost, cooling challenge, size challenge

10. The Communications Edge TM Satellite Communications Various system choices trading off # of satellites, radio complexity, system complexity –Geosynchronous: a few satellites a LONG way away –Low earth orbit: 1000 km orbits, 30-80 satellites Coverage is wonderful –Except inside any structure! Problem: capacity where you need it –Aggregate capacity of network several Gbps –HOWEVER: capacity/km 2 limited; can’t sell service to densely populated areas (where people who could pay for the cost live) To deliver 1 Mbps/square km from 1500 km up we need 10 Gbps and 1 degree beamwidth (1.5 meter antenna, can be built) Covers 10,000 square km; similar coverage 3000 small bstn, $140M capital cost BUT cellular basestations are granular: low entry cost, adjustable to localized demand, reconfigurable network Satellite universal communications is a proper sphere for government action: serves legitimate social need but not a viable business if self-funded

11. The Communications Edge TM SatCom RF Challenges Receiver: –Low-cost, high-sensitivity mm-wave receive easily achieved with outdoor directional antenna pointed at station –Not so easy with omnidirectional antenna: Omni collection area goes as square of wavelength At  = 2.5 cm (12 GHz), 24 dB penalty vs. 900 MHz cellphone with dipole –Assume 10 W at satellite 1500 km up, 1 degree beamwidth –Power delivered per cm: -99 dBm –Omni collection area 0.13 2 = 0.8 cm 2 ; power collected =-100 dBm –Noise floor at 1 MHz bandwidth =-114 dBm; if we need 5 dB (S/N) we have 9 dB margin Conclude: system may work outdoors but probably not indoors –Can fix by using large processing gain but wastes bandwidth

12. The Communications Edge TM End-result: heterogeneous network No one approach can or should support all requirements –UWB: in-home and short-range high-rate Standalone Integrate with fiber backhaul for ultrahigh datarate future network –WLAN: general-purpose high data rate in areas of high user density Standalone Integrate public hotspots with WAN for low-cost high-rate data –WAN: voice, essential / low rate data If you can use a handset you’re fine –SatCom: only the lonely (really need it) Integration with WAN not economically justified Universal connectivity is a public good, justifies public investment The real need (and opportunity) is interworking software so the user doesn’t know or care how the data is shipped –Loose integration via Internet is adequate for most purposes