1 Gb/s over 4-pair UTP-5 - Preliminary Simulation Results

Sailesh K. Rao
SDE Inc.
March 12, 1996.

The Split Passband Proposal
o	Allows one to pack more bits over low frequency band
o	Ease of creating and multiplexing two frequency separated signals
o	Avoids Baseline Wander problems
o	[diagram of proposed transmit spectrum]

Block Diagram of Proposed Transmitter
o	[diagram of Block Diagram]

9-level Lower Band Signalling
o	[diagram of 9-level signalling with probability of occurence]
o	"Extra" point used for synchronization between 4-pairs, SFD,
 	EFD etc. just as in 100Base-T2.
o	SNR penalty for 9-level vs. 8-level = 0.2dB

B-o-t-E SNR Calculations
o	The attenuation of 100m UTP-5 is specified to be
	Attenuation(f) < 1.967sqrt(f) + 0.023f + 0.05/sqrt(f) dB,  0.772<f<100
o	The NEXT Loss is specified to be
	NEXT(f) > 64.0 - 15.0log(f/0.772) dB,	0.772<f<100
o	SNR for lower band signal:
	At center frequency of 15MHz, SNR = 36.8 dB from above equations
	                              SNR = 32.4 dB from ISO/IEC-11801 Annex E,
	                                            assuming worst-case connectors
	SNR needed for BER of 10^-10 = 29.7 dB
o	SNR for upper band signal:
	At center frequency of 65MHz, SNR = 17.8 dB from above equations
	                              SNR = 12.0 dB from ISO/IEC-11801 Annex E,
	                                            assuming worst-case connectors
	SNR needed for BER of 10^-10 = 16.2 dB
o	Need more Excess Bandwidth for upper-band signal!

Transmit Pulse Shaping Filters
o	Lower-band: square root raised cosine, 20% excess bandwidth, 15MHz CF, 240ns
	[Diagram of Lower-Band Pulse shaping]
o	Upper-band: square root raised cosine, 40% excess bandwidth, 65MHz CF, 240ns
	[Diagram of Upper-Band Pulse shaping]

Transmitter Power Spectral Density
o	[Diagram of Transmitter PSD]

Received Power Spectral Density after 100m UTP-5
o	[Diagram of Receiver PSD and Worst-Case NEXT]

Example Analog Receiver
o	[Block Diagram of Example Receiver]

Preliminary Simulation Results - Genie Approach
o	Pretend Genie reveals optimum matched filters for given channel
o	[Diagram of great-looking 2-D constellations]

Conclusions and NEXT steps
o	Established Feasibility of Split Passband Approach for 1Gb/s over
	4-pair UTP-5
o	Put genie back in bottle - determine optimum analog/DSP receiver
	configuration
o	Study latency issues
o	Signalling Details
o	Echo Cancellation for Full Duplex operation