I am not sure what you have in mind with your question, but I believe that many people think that the achievable data rate (or goodput) is the appropriate performance measure of a point-to-point system, at least for modern communication systems that applies channel coding and ARQ to take care of errors in the transmission.

Multi-user systems are much more difficult to assess. It is quite common (in my area of physical layer research) that people try to maximize the weighted sum of the data rates of each user. The weights are either assumed fixed or updated to maintain a certain type of long-term user fairness (e.g., so-called proportional fairness). However, the multi-user performance can be measured in many other ways (it is actually a multi-objective optimization problem) and the selection of performance measure has a strong impact on how easy or difficult it is to perform radio resource allocation that maximizes the selected measure.

The multi-user case is desrcibed in more detail in my book "Optimal Resource Allocation in Coordinated Multi-Cell Systems": http://kth.diva-portal.org/smash/get/diva2:608533/FULLTEXT01

My question is caused by a necessity to understand modern methods of the communication systems (CS) performance analysis and optimisation. Below, I'll is not easy and confined a question by point to point (P2P) CS as the basic elements of PHY layer. Their performance determines performance of the PHY layer and network as a whole.

Unfortunately, I met difficulties in a search of materials on the methods of P2P systems optimization, as well as of the commonly used criterions of the systems performance assessment and comparison. Majority of authors consider, as the basic characteristic, the bit rate. Many authors use the power (energy per bit) or bandwidth (bit/s/Hz) efficiency of transmission, or joint P-B efficiency. More advanced researches are concentrated on tradeoffs between the bit-rate (or P-B efficiency) and methods of modulation and coding. BER is sometimes studied independently or is considered as an additional constraint.

The bit-rate is really a good measure, but how can we assess its closeness to the capacity of the system? Capacity of AWGN channel is determined by its bandwidth and output SNR. But I did not find any assessments of the capacity of CS as a whole, which take into account characteristics of digitizing and coding units. Only heuristic evaluations of the bit rate at the CS output.

I could omit these works, but I also did not find the results connecting capacity of the systems (P2P, with AWGM channel) and BER. The same concerns changes of CS bit rate and its limit values if the channel SNR is less than threshold and number of errors exceeds the correcting capabilities of decoders.

Some years ago, I've interested by optimisation of analog CS with feedback (AFCS) using original approach to optimization of estimation systems with noisy feedback and adaptively adjusted analog part (e.g. sensors, receivers, other units).

AFCS allow analytical formulation of MSE and optimization task, as well as its solution by methods of Bayesian estimation theory. Solution of the task gave the expressions for optimal transmission-reception and low boundary for MSE. Algorithm permits to design AFCS attaining the MMSE for each moment of the signal transmission.

(I’ll accent that principle of transmission has nothing common with ARQ. Each sample is transmitted independently and iteratively in definite number of cycles).

Analysis of solution showed that optimal AFCS transmit the signals with a bit-rate equal to the capacity of forward channel, in full concordance with known Shannon’s result for FCS. The same concerns P-B efficiency. There also appear another, not studied before effects briefly discussed in my paper (IEEE CL, v. 16, no. 5, 2012). If necessary, additional information can be found in my profile and answers to the questions of dr. Zekry in Research Gates (“The Shannon capacity limit of a communication channel, how to approach it practically?).

The feedback confines the field of optimal AFCS applications by short-distances and relatively low frequency signals. Lack of digitizing and coding units (lower energy consumption, costs and sizes) together with maximal P-B efficiency makes these systems useful for “one mile zone” communications (but not between the computers). For this reason, we began the researches in improvement of PHY layer of wireless sensor networks (WSN).

However, as a while, our probes to compare the characteristics of AFCS and any of known wireless sensors have failed.

AFCS transmitters are loss less (assuming internal noises are minimal), and transmit the signals without any added redundancy.

To compare the efficiency of AFCS and known link "sensor-base station" used in the same conditions, we should know – which part of power of the sensor transmitter is lost on transmission of the correcting bits. (they appear at least twice at the level of the packages forming and in PHY layer transmitters).

Not less complex is assessment of the number of signing bits at the receiver of the base station output, as well as of BER.

Briefly speaking: we search the method of adequate evaluation of closeness of the WSN node transmitters performance to theoretical limit. I would be pleased if somebody could help us although partially. This could result in common publication. I cannot exclude that this task was ever considered.

The better we'll organize the PHY layer, the more efficient network we'll get.

All is in order. The experienced in applied wireless colleague form Warsaw prompt me a monograph which I omitted in the wave of literature in the field. It answers to major part of my questions and doubts. This is interesting book: N. Hunn, Essentials of Short-Range Wireless, Cambridge Univ. Press, 2010. Thank you all who lost some minutes to think on my question.

Good to hear that you found what you were looking for!

I wanted to read the COMML that you referred to before replying to your previous message, but I haven't found the time to do that yet. However, it is in my pile of printouts...

Thank you. I've found a confirmation of great difficulties in this field, and interesting heuristic approaches to detour them. With regards...