Its difficult to find any situation where analog CS can do better because after all it has to be integrated / interfaced with digital device/s.

Thank you for the answers.

I looked through the links sent by dr Subramanyam and do not want to comment them in details. Only one remark: the presented there material is written very superficially. In many places information is erroneous.

The only benefit is a probe of systematic comparison of digital and analog communication systems (CS).

Blast of enthusiasm in plural comments to http://www.diffen.com/difference/Analog_vs_Digital shows a great need and interest to the topic described partially or scattered in more reliable publications.

I have no possibility to correct material of these links, and would recommend the young scientists to make this using verified sources (e.g. books of S. Haykin, B. Sklar – they are written at level accessible for engineers). This is worth to do.

Concerning points 1-4 of dr Subramanyam text: they concern only small part of the question. I'll some develop these points.

Main differences in functioning of ACS andDCS.

All the signals generated by physical objects and medium are analog. On the non-quantum level, all the sources of discrete signals are thought out by people. Digitizing the signals has appeared due to appearance of computers. Analog signals not only were to be delivered to computers but also in the form enabling their processing.

Digital CS make this at once, in input units of transmitters which form and emit sequences of binary signals prepared for digital processing in receivers.

Digitizing consists in sampling and quantization of the signals in ADC. Signals at ADC output are binary words which represent the numbers of quants referring to the values of input analog signal.

That is important: quantization is lossy operation and causes irremovable errors in the signals restored by receivers. Also, omitting details, I will note that digital CS do not allow any analytical assessment of the final accuracy of restored signals.

Analog CS transmit the analog signals directly, without digitization and coding, only using chosen type of modulation.

The received signal (naturally, distorted by channel noise) is demodulated and routed to addressee directly (in earlier CS) or (in later of CS ) after digitizing and filtered by noise suppression algorithms.

That is important: unlike digital CS, analog CS transmit the signals without quantisation errors and allow analytical assessment of the accuracy of received signals (i.e. computation of mean square error of transmission - MSE) .

What further?

I completely agree with colleagues: development of coding and digital theory and technologies, and efficient solution of basic communication tasks by digital CS proved their full superiority over known analog CS.

Only one question remains – is this final and full winning? Or there are the fields where analog CS may work better and be cheaper than digital ones?

I asked this question, because all pioneer works in analog CS with feedback (AFCS) clearly showed - this is possible. Why I think so: unlike digital CS, AFCS permit to construct full mathematical model of transmission–receiving process. Using this model, one may determine MSE and requirements to the parameters of transmitter and (digital) receiver which minimize MSE for each moment of the signal transmission.

AFCS designed according to these results would achieve the upper boundary of performance (Shannon’s limits) without coding. The latter means full and optimal utilisation of all resources of the transmitters and receivers of AFCS including the channel bandwidth.

On the other hand, this cannot be done in digital CS. For this cause, all of them utilize their resources not completely and not optimally. This permits to assume that there should be fields where “analog vs. digit” is better. Can and how can they be found?

I think, if pioneer researches were continued else 3-5 years, modern communications would be some another. But something ceased these researches. What? We may think together.

-------

The said above partially concerns the answer of dr Ahmad. - It is really not easy to find situation where analog CS can do better. But I cannot agree that the reason is their integration with digital devices. Later versions of analog CS also used digital devices in receivers but lost.

With regards

Anatoliy Platonov.

I'll explain the reason of my question: some years ago I've published first results confirming positive answer on the question, and continue the researches. It was interesting to know - did somebody touch this topic? As a while, situation is unclear. Closing the question, I will add some remarks:

1. All recently known "one-mile zone" transmitters of signals from analog sources (sensors of WSN nodes, microphones, earphones....) are integrated with the sources. Formed by them signals are immediately digitized and transmitted as the coded binary sequences.

2. Our researches show that it is much more efficient to transfer digitising the signals to the receiver of base station. Origin analog signals should be transmitted by properly designed loseless analog modulators adjusted via feedback channel. Digitizing should concern the demodulated signals with further processing of their binary codes by corresponding algorithms computing the estimates of origin analog signal.

3. Analog feedback CS (AFCS) work according to this scheme. This allow analytical definition of the parameters of modulator and algorithms used for computing the estimates in the receivers, which minimise MSE of transmission. The obtained result - optimal transmission/ reception algorithm contain all basic information permitting to design optimal AFCS. So designed systems transmit the signals with a bit rate equal to the capacity of the systems, and their power-bandwidth efficiency attains the upper value.

4. Principle of transmission, multi-component criterion of performance, numerous trade offs and other difficulties make optimisation of digital CS (DCS) using MSE impossible. Therefore, all DCS are not optimal and not fully utilize the allotted power-bandwidth resources. This decreases their performance, and makes transmitters more complex, energy consuming and expensive, channel banwidth is utilised not optimally.

For this reasons, in comparison with DCS designed for the same applications, optimal AFCS may grant greater coverage, power - bandwidth efficiency, rate of reliable transmission, lower cost of transmitters and longer lifetime of batteries.

Conclusion: signals from the analog sources should be transmitted by the analog FCS. This will be more efficient and cheaper than their digital transmission that is especially important in WSN and other short-range applications.

Thank you all who looked at this question.

Dear colleagues

Dear Anatoliy,

This question has been asked by some researchers on the researchgate from different perspectives and various motivations. I my self put few questions around analog versus digital. In a previous question for me about the possibility of the practical realization of the shanon channel capacity, you introduced the AFCS as a communication system that can reach the shanon limit while the digital communication systems can not basically achieve this limit because of the quantization losses may reduce the correct detection capability of the signal at the receiver. Because of the quantization process is an approximation process.

I will not repeat the advantages and disadvantages of the analog and digital systems. The matter of fact, the systems are digital. In the sense that the core of the system is digital while the analog signals is displaced to the shell or at the peripheral of system. I think, The evolution theory is applicable also on the electronic system developments. It is the enablers of system implementations which stand beside the binary digital systems to master the system nature. That is the technology played a major role for favoring the digital versus the analog especially in the processing. Another very important factor is the existence of digital electronic memory which is vital for the program ability and intelligence of the digital systems. More over the formality of the digital systems in the design and testing. This means that they are easier to build .Digital circuits are well regularly structured such that their integration density will be much higher than the analog circuits.

The advancement in the integrated circuits towards very large scale integration with ten millions of scaled transistors operating at low power supply voltage. in a chip fabricated simultaneously reduced the energy consumption per operation.

When advancing the digital systems, the scientists and engineers worked to treat the the drawback of the digital systems.

A the same time they could develop very powerful data converters. The sampling speed is in the order of hundred megahertz with acceptable resolution.

Concerning the quantization error, practically it must be made appreciably smaller than the noise and distortion error. In this way, it will have a negligible effect on the error performance of the system.

In addition, Source coding and channel coding played a fundamental role to make the digital transmission and storage affordable and cost effective. Source coding may be lossy but the losses are kept within acceptable limits for the users.

They play on the concept : It is good so long the customer is satisfied.

I think the objective of the electronic system builders is not to produce ideal systems but satisfactory operating systems.

Finally, the presence of a digital core makes the piece of equipment performing many other functions than communicating.

Please look at your smart phone!

Analog and digital will coexist side by side and they will be further evolved and their architecture will be favor a hybrid system. Anatoliy, even your system is hybrid.

It is a long comment. Sorry

Thank you.

Dear Abdelhalim,

Thank you for interesting and wise comment. I fully agree with your assessment of situation. Digital CS (DCS) will be good so long the customer is satisfied. Moreover, they always will be “grand sister” of analog communications as a single method of inter-computer and long distance communications, and capable to transmit high frequency signals.

The main limitation of analog CS (AFCS) is the time delay in adjusting the transmitters via feedback. This determines the field, where they will be substantially more efficient than DCS, is short range transmission of the relatively low frequency analog signals.

It was a small tragedy of communications when, in the middle of 70th , researches in AFCS were ceased. As I already wrote, yet 3-5 years of work at the same level of intensity could give the solutionwhich I found thirty years later. Practically, all was already done. I only took into account possible saturation of the transmitters. If this was done in 70th , communication world would be much more harmonic. Analog and digit really would work side by side.

Digital CS went into “one mile zone”, because there were no other way - AFCS theory had no applied results. It was natural to use “easy to build” communication schemes permitting relatively low cost realisation in digital technologies. The main goal was to solve the task, and less important – cost of solution. Nowadays, situation has changed. I’ll quote some of opinions expressed in recent publications:

“Next generation of high performance wireless networks must support increase in data rate, better coverage, spectral efficiency and higher reliability. In addition, many applications… need extreme energy efficiency, as well as new wireless networking paradigms” [Goldsmith, A., “The next wave in wireless technology: challenges and solutions”, Keynote presentation, IEEE Int. Conf. on Communications, Computing and Networks, ICCCN’2009, San Francisco (2009)].

“Whether research at the physical layer (PHY) of networks is still relevant to the field of wireless communications? We exemplify areas where any improvements are expected to be marginal. We then discuss whether the research directions taken in the past have always been the right choice and how lessons learned could influence future policy decisions”. [Dohler, M., Heath, R.W., Lozano, A., Papadias, C.B., Valecuela, R.A., “Is the PHY layer dead?”, IEEE Comm. Magazine,. 49(4), 159-165 (2011)]

“The ability to send data from receiver back to transmitter opens new doors in the way of modulation schemes”, [Chance, Z., Love, D. J., “A noisy feedback encoding scheme for the Gaussian channel”, Proc. IEEE Int. Conf. on Acoustics Speech and Sign. Processing, ICASSP’2010, 3482-3485 (2010)].

These and other opinions show growing need in improvement of short range CS performance and efficiency, and search of possible solutions. It is some strange, but AFCS permitting to solve many of formulated above tasks look as too outrunning and not required. Nevertheless, I also believe in evolution and AFCS. Ever they will be claimed.

I accept all basic thoughts of your comment, and will allow myself only small remarks clarifying some of possibilities of AFCS and role of optimisation.

From your text::

“…digital communication systems cannot basically achieve this (Shannon’s) limit because of the quantization losses may reduce the correct detection capability of the signal at the receiver”.

“… objective of the electronic system builders is not to produce ideal systems but satisfactory operating systems

”…, the presence of a digital core makes the piece of equipment performing many other functions than communicating.

Answers:

Quantisation errors are not essential – they can be reduced by good ADC. The main source of transmission errors are forward – feedback channel noises which crucially decrease SNR at the receiver input. None DCS may suppress their influence in the way minimizing MSE and transmit the signals with a bit rate equal to the capacity of the system. Optimal AFCS may.

To the second sentence: Achievement of information and accuracy limits permits to increase the coverage and to shorten the time necessary for transmission of given amount of information (e.g. data from "awaken" sensors). Presence of ADC and coding units (non forgetting modulator), even realized in nano-technologies, increase the cost, energy consumption and sizes of transmitters in comparison with optimal AFCS transmitters (only modulators) realized in the same technologies. I also will emphasize that the objective of adequately performed optimisation is to give the designers analytical backgrounds for designing the systems with better characteristics than every satisfactory operating system designed in another way.

To the third sentence: Yes, digital core enables performing many other functions than communicating. But many of these functions can be realised by digital core of base station via feedback. Why should we overload the battery supplied small transmitter? There also exists a large number of CS which do not need additional functions from their transmitters (e.g. wireless microphones). Mash topology networks also can be organised at the level of base stations (master nodes).

Thank you once more for interesting answer and thoughts.

I also regret for a long comment. Sorry. This is influence of Sunday.

All the best!

P.S. I do not touch the role of memory in communications. I did not work with it and related tasks. But if it existed, I would give Nobel prize to its creators.

There is not much I can contribute to this interesting discussion, but I remember hearing Tor Aulin (IEEE Fellow) saying he wrote some papers comparing analog and digital communication systems. And, last year I heard a very interesting talk concerning hybrid systems, which can get lower to the Shannon capacity bound:

Ruengeler and Vary: "Surpassing Purely Digital Transmission: A Simplified Design of Hybrid Digital Analog Codes"

http://www.ind.rwth-aachen.de/fileadmin/publications/ruengeler13.pdf

Thank you, Colleague Geiger, for mobilizing me to continue a discussion. The work currently absorbed me and some confined possibilities of its continuation.

Thank you also for the new names and publications. My searches in literature (on the analog CS optimization and design) had failed. The world is digital and "mad" scientists and engineers interested and working in analog CS are almost not seen. Or they do not publish the results or are not published by the central journals. Only top-level theorems of existence in IEEE Inf. Theory but without any interest to the practice.

Now, I got the grant permitting my group to design and study the prototype of new AFCS. Theory also gives new outlook on the known and yet not studied problems.

I will turn back to RG discussions. It is very interesting and important to read the points of view ("fresh look") of colleagues. As well as to get information which I could not find myself (partially expressed in my questions in RG).

With regards

Anatoliy Platonov