0
\$\begingroup\$

I was reading a textbook, Data Communications And Networking, by Behrouz A. Forouzan, which says:

A digital signal is superior to an analog signal. The tendency today is to change an analog signal to digital data.

I don't quite get it, how is a digital signal superior to an analog signal? What's the benefits of changing an analog signal to digital data?

\$\endgroup\$
  • 11
    \$\begingroup\$ This textbook sounds pretty bad. Maybe stop using it and use another one. \$\endgroup\$ – DKNguyen Jun 16 at 15:24
  • 2
    \$\begingroup\$ yeah, that textbook is simply wrong. Wrong from an engineering point of view (see Aaron's answer). Also, from an information-theoretical point of view (there's the whole field of rate-distortion theory that describes how much worse a digital signal is compared to the original analog one, in terms of transporting information). Can we do cool things with digital signals that we can't do with analog signals? Indeed, we can. Does that allow us to build better digital systems than we can build analog systems for particular use cases? yes. Is it right that "a digital signal is superior"? Hell,no \$\endgroup\$ – Marcus Müller Jun 16 at 17:07
  • \$\begingroup\$ You shouldn’t tell that the textbook is simply wrong (Just by judging a line). Networking books focuses on long distance communication. Plus it needs many level encryption, exchanging data between thousands of storage devices. \$\endgroup\$ – Sadat Rafi Jun 17 at 10:12
7
\$\begingroup\$

As with most things in engineering, it depends on the trade offs. If you want to add complexity in order to gain noise immunity, then yes the book is correct.

However, maybe you have a circuit where the analog source is only 0.1" away from the A2D, would I put a PCM circuit for that? No!

Maybe you have an analog sensor that requires an op-amp for some reason, now you have a nice low impedance driver, which can withstand noise better, so again, PCM isn't needed.

There are more applications for PCM than what I've described check Wikipedia: Wikipedia on Pulse-code_modulation

To specifically address the questions:

how is a digital signal superior to an analog signal?

This typically has to do with noise immunity. An analog signal with 100mV of noise on it is most likely not very usable. However a digital signal with 100mV of noise on it will work just fine. Look at the Vih and Vil of almost any digital chip as a starting point.

What's the benefits of changing an analog signal to digital data?

This usually has to do with a storage requirement like in CDs(check the Wikipedia link), or transmitting over long distances or on a medium that only supports digital signals. In the later two cases, it really boils back down to noise immunity.

| improve this answer | |
\$\endgroup\$
3
\$\begingroup\$

For long-distance communications, it's easier to amplify a digital signal, because you don't have to worry about issues like linearity and noise (to a point). You can also apply error detection and correction to digital data.

Of course, that's assuming you're willing to live with the limitations in terms of bandwidth (dictated by the sample rate) and resolution (dictated by the quantization).

Does the textbook not cover these points? If you want a more specific answer, we need more context. What is the specific book you're looking at?

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Data Communications And Networking by Behrouz A. Forouzan \$\endgroup\$ – amjad Jun 17 at 8:20
1
\$\begingroup\$

This is a big topic and you'll find plenty of existing text on the subject on the internet.

On aspect is noise rejection. Digital signals benefit from relative ease of noise rejection over an analogue waveform. In an analogue waveform, it is not readily apparent what is the wanted part (signal) and what is the unwanted part (noise). With a digital signal, a noisy signal can be more easily restored to a true logic voltage level.

If the application warrants it, the digital signal bandwidth allows it and the latency is acceptable, then error check and correction (ECC) algorithms can be applied to blocks of the digital data to allow noise disrupting numbers of bits to be corrected. This improves noise rejection yet further.

Digital signal processing is a subject in itself and there are many advantages, and many disadvantages such as quantisation errors/noise, higher bandwidth requirements etc. Again, this is something to investigate further for yourself.

| improve this answer | |
\$\endgroup\$
1
\$\begingroup\$

Nation security drove the move from analog telemetry, to digital telemetry.

Analog telemetry could not achieve 50dB SNR (about the point where the width of the INK PLOTTER ink line set the resolution) at the distances needed to monitor the (mis)behaviors of the missile systems.

By using Analog_Digital_Converters, the data_link limitations were greatly altered. If your rocket mis-behaved in mid-trajectory, you had a better chance of diagnosing the cause, because you had better reporting of the minute variations of the fluids and the vibrations and the temperatures.

I remember my first Chief Engineer staying late, several days in a row out in the lab. I went over, curious, and he explained "This is a 10-bit ADC from the Saturn Instrumentation Segment. Has a problem. Since I designed it, 10 years ago, I'm the best person to diagnose it and then ensure its properly calibrated after the NASA_spec repair_people replace the failure."

The ADC was a 10_bit current_steering successive_approximation circuit, with 10 current sources, and at least 10 discrete FlipFlops to implement the binary-search algorithm. Plus 10-bit output data holding register.

I recall a Cordwood Module FlipFlop of size 1cm by 1cm by 2cm, using capacitive input coupling to create the toggle-pulse behavior with diode steering logic and 2 transistors of 2N706. When Fairchild brought out their UL923 logic, the size was about (1/2 cm) cube, but used enough (on_chip) transistors to function down to DC without re_design.

All in 8" by 8" by 24" chassis.

By the way, in trying to develop (debug, stop explosions from broken piping) their Soviet N1 moon-launch rocket in the 1960s, with their 13 engines needing lots of large piping for oxidizer and propellant with shared turbine pumps, the Soviets kept adding more and more telemetry channels as each version rose from the pad ----- and then exploded at some distance aloft.

The final version (#5) had 13,000 telemetry channels and over a GigaBit of downlink data streaming.

After that 5th explosion, still not diagnosable (too much vibration in long piping?), the Soviets gave up their moon efforts.

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.