DIGITAL VOICE NETWORKS

1. DIGITAL VOICE NETWORKS ECE 421E Tuesday, October 02, 2012

2. ADVANTAGES OF DIGITAL COMMUNICATION NETWORKS • Ease of multiplexing • Ease of Signalling • Use of modern technology • Integration of transmission and switching • Signal Regeneration • Advanced Performance Monitoring • Ability to integrate other services • Ability to operate at low Signal to Noise Ratio • Ease of encryption

3. COST OF MULTIPLEXING • The Cost of TDM systems is much lower that that of analogue multiplex systems, e.g. FDM.

4. SIGNALLING • Digital systems allow control information to be inserted into and extracted from a message stream independent of the mode of transmission. • Signalling equipment are designed separate from transmission systems allowing control functions and formats to be modified independently.

5. USE OF MODERN TECHNOLOGY • Multiplexer and switching matrix for digital systems are implemented with the same basic circuits used to build computers • Special LSI Circuits have been developed specifically for telecommunication functions e.g Voice Codecs, Multiplexing, DSPs, etc. • Low-cost of digital circuitry allow for implementations that would be very expensive if developed on analogue platforms, e.g large non-blocking exchanges. • Digital technology provides easier and cheaper interfaces to fibre-optic cable systems.

6. INTEGRATION OF TRANSMISSION AND SWITCHING`

7. BENEFITS OF FULLY INTEGRATED DIGITAL NETWORKS • Long-distance and local voice quality are identical in terms of noise, signal level, and distortion. • Since digital circuits are inherently four-wire, network-generated echoes are eliminated, and true full-duplex, four-wire digital circuits are available. • Cable entrance requirements and mainframe distribution of wire pairs is greatly reduced because all trunks are implemented as sub-channels of a TDM signal.

8. Signal Regeneration PCM Signals can be regenerated at suitable intervals

9. SPEECH DIGITIZATION ELC421

10. SPEECH FREQUENCY RANGE • The range of frequencies that the human ear can perceive - 20Hz – 20KHz (Natural voice frequency range) • Acceptable level of intelligibility is obtained by transmitting voice in range 0.3 -3.4 KHz • Most of the voice energy is in this band.

11. SAMPLING • Nyquist Criterion/Theorem • Fs > 2fmax where fmax is the highest frequency in the analog input signal

12. PULSE AMPLITUDE MODULATION (PAM) SPECTRUM

13. ALIASING/FOLD-OVER DISTORTION • Occurs when fs < 2fmax resulting in an overlap of the spectrum. • Aliasing/fold-over distortion is avoided in telephony by band-limiting the signal to the range 0.3-3.4 KHz. • By over-sampling at 8KHz, the sampled signal is sufficiently attenuated at the overlap frequency of 4KHz

14. EXAMPLE OF ALIASING DISTORTION

15. END-TO-END PAM SYSTEM The response of the reconstructive filter is modified to account for the spectrum of the wider staircase samples.

16. BAND-LIMITING FILTER DESIGNED TO MEET ITU-T RECOMMENDATION FOR PCM VOICE CODERS • ITU Standard requires14 dB attenuation is provided at 4 KHz.

17. SAMPLING TECHNIQUES

18. PULSE CODE MODULATION (2)

19. PULSE CODE MODULATION (2) • A/D Converters and D/A Converters are inserted in a PAM system discussed above • All sample values a quantization interval are represented by one discrete value creating quantization noise

20. PCM SYSTEM FOR SPEECH COMMUNICATION

21. QUANTIZATION ERROR AS A FUNCTION OF AMPLITUDE Quantization Error can be expressed as ɛ = V – Vq

22. LINEAR QUANTISATION • Assume that the probability distribution of error is constant within the range ±S/2. • Average quantisation noise is given by: • Where µ = mean • The range of the quantisation error, i.e ±S/2 determines the limits of integration

23. AVERAGE QUANTIZATION NOISE OUTPUT POWER FOR LINEAR QUANTISATION

24. SIGNAL TO QUANTISATION NOISE RATIO (SQR) SQR is a measure of the performance of a PCM system. • Where • E{..} is the expectation or average • x(t) is the analog input signal • y(t) is decoded output signal • Assumptions • Error y(t) – x(t) is limited in amplitude to S/2 where S is the height of the quantisation interval • Sample value is equally likely to fall anywhere in the quantisation interval • Signal amplitude is confined to the maximum range of the coder

25. SIGNAL TO QUANTISATION NOISE RATIO (SQR) • If all quantisation intervals have equal length, i.e uniform quantisation, the quantisation noise is independent of the sample values and SQR can be determined as: Where v is the rms value of the input and q is the quantization noise.

26. SQR FOR SINEWAVE • For a sinewave with Peak Amplitude A, SQR is given by:

27. EXAMPLE • A sinewave with a 1-V maximum amplitude is to be digitized with a minimum SQR of 30dB. How many uniformly spaced quantization intervals are needed and how many bits are needed to encode each sample.

28. SOLUTION 1. 2. • Number of quantization Intervals is 1/0.078 = 13 of each polarity yielding 26 intervals. • Number of bits N = log2(26) = 4.7 or approximately 5 bits per sample

29. IDLE NOISE CHANNEL • Noise occurring during speech pauses is more objectionable than noise occurring during speech. • Idle Noise power is usually specified in absolute terms separate from quantization noise. Typical figure is 23dBrnCO.

30. DIGITAL SWITCHING ECE 421

31. SWITCHING PRINCIPLE

32. PCM MULTIPLEXING PRINCIPLE

33. PCM TRANSMISSION SYSTEM