<< Chapter < Page Chapter >> Page >
L t = α t f i n Δ f t ,

where α t is determined by the degree of stopband rejection desired Specifically, a t = 0 . 22 + 0 . 0366 · S B R , where SBR is the minimum stopband rejection in decibels. A typical value for SBR is 60 dB, yielding an α t of 2.42. and Δ f t is the tuner's transition band. In this case, the transition band can be no greater than the difference between B t and f s . If we assume the use of this limiting value, L t is given by

L t = α t f i n f i n M t - B t

Substituting this expression and expressing all sampling rates in terms of the input rate f i n produces an equation for the total number of multiply-adds required.

G total = 2 f i n · { 1 + α t f i n f i n - M t B t } + 2 f i n N M M t { Q + log 2 N } = 2 f i n · { 1 + α t 1 - ( M t B t f i n ) + N M M t { Q + l o g 2 N } }

Another useful form of this equation makes the functional dependence on f s more explicit. We do this by using the expressions

N f s Δ f , M t f i n f s , and K N M

and the assumption that a radix-2 FFT is employed to compute the DFT. With these, the expression for the total number of multiply-adds can be written as

G total = 2 f i n + 2 α t f i n 1 - B t f s + 2 f s K · { Q + l o g 2 f s Δ f }

Parameter optimization

Given expressions such as those shown in [link] and [link] it is possible to accurately estimate the total amount of multiply-add computation needed for a tuner/transmux processor. It is also possible to perform tradeoffs between the various parameters in order to optimize the resulting design. While this can in principle be done with any of the design parameters, we demonstrate in this section the computational implications of varying the parameter f s , the input sampling rate to the transmultiplexer. In practice, this usually turns out to be one of the designer's most important parameter choices.

[link] shows the computational requirements for a hypothetical transmultiplexer. In this case, the input sampling rate f i n is assumed to be 6.4 MHz. The tuner must select an FDM telephone supergroup from the input signal and demultiplex all 60 voice channels in the supergroup. The tuner's bandwidth B t must therefore be greater than or equal to 240 kHz and f s must exceed that. For the telephone demultiplexing application, the channel spacing Δ f is usually 4 kHz and the over-sampling factor K is typically chosen to be unity. [link] shows five curves, one for each segment of the computation and one for the composite. The number of multiply-adds required by the input mixer is constant, since the input sampling rate f i n is fixed. The computation required by the tuner's filter falls as f s rises from 240 kHz and tends toward the input Nyquist frequency of 3.2 MHz. The cause of this can be ascertained by examining [link] . As f s decreases toward B t , the transition band decreases, L t increases hyperbolically, and the amount of computation needed for the tuner's filter grows without bound.

Figure two is a graph, with horizontal axis labeled Intermediate Sampling Rate f_s (in kHz), and vertical axis labeled Mega-multiples/Seconds. The horizontal axis ranges in value from 0 to 1000 in increments of 200, and the vertical axis ranges in value from 0 to 150 in increments of 50. There are six curves on this graph. Beginning from the origin are two lines with a constant, extremely shallow slope so that their endpoints are approximately (1000, 15) and (1000, 20) respectively. There is also a roughly horizontal line along the vertical value of approximately 12. The end points of these lines are labeled Weighting FFT Mixer There is a vertical line along a horizontal value of approximately 240, and its distance from the vertical axis is labeled Tuner Bandwidth (One Supergroup). A line begins at approximately (300, 160) with a strong negative slope, until approximately (400, 70) where the slope becomes much more shallow, and the curve continues to decrease at a much slower rate to a final point of (1000, 45), labeled Tuner. This includes a couple black dots along its path, at approximately (320, 110), (390, 75), (500, 55), (550, 52), (600, 50), and (1000, 45). A dashed curve begins slightly above the aformentioned curve, follows closely to its shape until the three dots that are closely-spaced together on the shallow part of the line. On the dashed curve, these three points are labeled, Most Efficient Choices. After these points, the curve begins to increase with a shallows slope to an end point at (1000, 80) labeled Total. There is a caption box with the following equations: f_s = 6.40 MHz, α = 2.5, B = 252 kHz, Q = 10, ∆f = 4 kHz.
Tradeoff between the Design Parameter f s and Total Computation in a Hypothetical Supergroup Transmultiplexer

The next two curves describe the effect of f s on the two components of the transmultiplexer. For a given value of Q , the computation required by the preprocessor is strictly proportional to f s . The FFT's computation rises slightly faster than proportionally since the number of FFT bins grows as f s does. The sum of these constituent curves represents the total amount of multiply-add computation needed. Note that it has a broad minimum. It rises precipitously as f s decreases toward B t and more slowly as f s increases toward its other limit f i n 2 .

Get Jobilize Job Search Mobile App in your pocket Now!

Get it on Google Play Download on the App Store Now




Source:  OpenStax, An introduction to the fdm-tdm digital transmultiplexer. OpenStax CNX. Nov 16, 2010 Download for free at http://cnx.org/content/col11165/1.2
Google Play and the Google Play logo are trademarks of Google Inc.

Notification Switch

Would you like to follow the 'An introduction to the fdm-tdm digital transmultiplexer' conversation and receive update notifications?

Ask