1. Sept. 27, 2006 Lab #3 Due Today Tie it all together: Bandwidth, bandlimiting, information capacity example EM Spectrum Noise

2. 802.11 Bandwidth 802.11b and g use 2.4GHz band They have 14 channels with 5MHz spacing –Bandlimiting – at +/-11MHz, signal must be 30dB down. At +/- 22MHz, signal must be 50dB down. –Lots of overlap between channels, requires good network design

3. 802.11 WLAN Example Dumb design: co-locate 3 APs using 802.11 channels 1, 2, 3 Peak power for each channel at center frequency (f c ) of channel Power drops moving away from f c but is not 0 when next channel is reached

5. 802.11 Example – Issues With WLAN Channels overlap SNR is low If co-located, even channels farther apart in frequency domain will overlap Noise from other sources will decrease SNR even more (microwave ovens, bluetooth, cordless phones)

6. What we can do Move APs apart to reduce interference Channels adjacent in space should not be adjacent in frequency –If using 3 APs, use channels 1,6,11 –If using 3 APs, use channels 1,4,8,11 Noise will go down, SNR will go up, information capacity will go up

7. Electromagnetic Spectrum range of all frequencies of electromagnetic radiation broken into subranges EM Spectrum - physical characteristic Spectrum allocation - humans dividing spectrum into different uses and designating who can do what

8. EM Spectrum Divisions www.ntia.doc.gov/osmhome/allochrt.pdf

9. Frequency Allocations Allocations in Canada by Industry Canada, and in US by FCC History of EM Spectrum Allocation: –No allocations at birth of wireless communication –1912 first wireless communication control in US – easy to get bandwidth –1927, Federal Radio Commission given power to grant or deny radio access –Power to grant or deny access passed to Federal Communication Commission in 1934 –Granted spectrum based on “Best Public Interest” –In 1994, started to auction off frequencies in 1994

10. Frequency Spectrum Auctions Highest bidder obtains license granting holder right to emit EM at certain frequencies in a certain area –License generally granted for 8 years Supposedly done to create “free market” for EM spectrum. Fetches billions of dollars for FCC which is then put into general coffers

11. Frequency Spectrum Auction Criticisms How can FCC auction off something that is part of nature How can market price be established before there is a market Unlicensed bands are what has driven WiFi growth, unlicense more (or all) bands to drive more R&D New technologies could make EM spectrum for communication a “black box” (see next slide)

12. EM Spectrum as a Black Box Why do bands of spectrum have to be allocated for certain things? If you have data to transmit, why can’t we just transmit it without regard for which frequency it will actually go out on Therefore eliminate idea of allocation Don’t eliminate regulatory commissions – would still need “traffic cops” Would bring down barriers to entry, and create true “free market”

13. EM Spectrum as a Black Box (2) Basic technology exists to do this (frequency hopping/spread spectrum) Still would require a large advancement in technology Ideal: Enormous world-wide wireless network using all of RF band. –Don’t need to know which part of band you use, just need to know how to send/receive data Not there yet, radio commissions unlikely to budge on current state until technologies can support this

14. Noise

15. Noise Arch-enemy to signal Distorts analog signals, causes errors in digital signals Limits information capacity –I = 3.32 * BW * log(1 + SNR) –If noise = 0, SNR→∞, log(1+SNR)→∞, I→∞ –Data rate only limited by total amount of data you have if no noise

16. Uncorrelated Noise Noise from system not signal –Is present even if signal is not Sources External to system: –Atmosphere: electrical disturbances –outer space: cosmic radiation –man-made: electronics, motors, lights etc.

17. Internal Uncorrelated Noise Source of noise is internal to system Thermal noise –thermal movement of electrons. –Is constant across all frequencies. –Can be expressed mathematically: Noise (in watts) = KTB K=Boltzmann’s Constant = 1.38x10 -23 J/k T = temperature in Kelvins B = Bandwidth in hertz

18. Internal Uncorrelated Noise (2) Shot Noise –Due to random movement of electron flow in currents –electrons take different paths, sometimes faster, sometimes slower. This random flow causes shot noise.

19. Correlated Noise Noise related to or generated by signal Distortion –Sine wave is amplified too much to give a square wave –Harmonic Distortion

20. Signal to Noise Ratio (SNR) Measure of signal power with respect to noise power –Expressed as ratio, or in dB Most fundamental of communication system measurements –Gives first order estimate of how well system will perform Example: Signal strength is 1 watt, noise strength is.002 watts. What is SNR?

21. Noise Factor and Figure Measurements of how much a system distorts or changes the signal to noise ratio –Usually measures how much extra noise the system added Noise factor = SNR (in) /SNR (out) Noise Figure is just the Noise Factor expressed in dB –Noise Figure = 10 log (SNR (in) /SNR (out) )

22. Example A nonideal amplifier has the following parameters: –Input signal power = 2x10 -10 W –Input noise power = 2x10 -18 W –Power Gain = 1,000,000 –Internal Noise = 6x10 -12 W Determine Input SNR, Output SNR, noise factor, and noise figure