Spectrum is a precious resource, which encourages system design developments that improve bandwidth efficiency. This paper is the first to introduce the performance of the 1024-QAM signal implemented with the nonlinear switch-mode mixer modulator (SM$^{3}$) power circuit element. Measurements of modulation precision show less than 0.4% error-vector magnitude. Signal ACLR is -54 dB, peak output power is 2.5 watts, and the transmitter efficiency exceeds 50% under modulation and including all linearization.

Energy efficiency of wireless links is very important in modern communication systems. A survey of papers evaluating energy efficiency however shows that evaluation approaches are not consistent, resulting in calculations that are not comparable with each other. Indeed, even the units of the various energy efficiency evaluations are not identical. To address this problem a new model is developed here that takes into account the many dependencies among multiple wireless link parameters important to evaluating energy efficiency of the link. The unit of bits per joule is adopted for this new model. This new model is evaluated for five deployed wireless links that have very different parameters and operating objectives. The need to establish standardized cases to allow direct comparisons is identified and discussed.

While users of spectrum pay large sums of money for spectrum access, they do not sell spectrum to their customers but rather they market data rates. Developing radio technologies that maximize data rate within a finite spectrum slice is a very important goal to such spectrum access businesses. This requires radios to provide high values of spectrum efficiency (bits per second per Hertz-bandwidth (bps/Hz)). This paper is the first to introduce a conventional Nyquist filtered 4096-QAM signal implemented with a switching-circuit based transmitter. Even by having no circuit linearity in the RF path and using polar modulation, the reported signal quality measurements confirm less than 0.9% error-vector magnitude for this 12 bps/Hz constellation. Signal ACLR is -54dB, peak output power is 2.5 watts, and the transmitter efficiency exceeds 50% when modulated and including all linearization power. The limits of using EVM as a signal quality metric are presented and discussed.

The decision to use OFDM-based modulation for 5G-NR forces a low-valued ceiling on achievable energy efficiency from any linear power amplifier. The 5G network also has an objective to significantly increase its energy efficiency with respect to the energy draw of the LTE network. To get the desired efficiency improvement, at the transmitter, requires circuit nonlinearity. This paper introduces a transmitter design that accurately produces 5G-NR modulation while maintaining an operating energy efficiency exceeding 40%, well above the 14% of typical LTE linear transmitters. Circuit nonlinearity is maximized, using switch-based RF design, in order to maximize transmitter efficiency. Sampled signal processing is therefore required, and is used here within a polar modulation architecture. Measurements of this approach implemented using GaN HEMT technology, are provided which include the efficiency impact of crest factor reduction signal processing.

The 5G network has a goal to significantly increase its energy efficiency with respect to the energy draw of the LTE network. The decision to use OFDM-based modulation for 5G-NR forces a low-valued ceiling on achievable energy efficiency from any linear power amplifier. This ceiling is lowest for amplifiers operating at frequencies near and above 30 GHz: the millimeter-wave bands. Transistors behave differently at these high frequencies, and the modulation used must change to match to these different characteristics. Ring oriented constellations with polar filtering meet these changed conditions.

To maximize energy efficiency in communications, particularly wireless communications, it is necessary to minimize power dissipation in all circuitry. This joint minimization is unfortunately not independently achievable. Decisions on the type of signal modulation adopted for use in any particular application can and do put lower limits on achievable circuit power dissipation, and therefore putting a ceiling on achievable energy efficiency. Such restrictions must be removed to fully realize our Green Communications objectives. Here the reasons for the energy efficiency limits are presented, which results in a new Figure of Merit for appropriateness of a signal selection with respect to maximizing available communication energy efficiency.

While polar modulation is a transmitter technique that is known to maximize energy efficiency, it also has no circuit linearity and traditionally is unable to produce signals that contain envelope zeros such as QAM and LTE. This polar transmitter solves this weakness and is modulation agnostic across the decade-wide tuning bandwidth of 200-2500 MHz. In particular, a conventional Nyquist filtered 256QAM signal is generated with error vector magnitude (EVM) less than 1.5% across the frequency range, and is below 1% across the bottom decade of frequency - without use of digital predistortion (DPD) linearization. Peak output power is 2.5 watts.

Implementing any massive MIMO array involves managing transmitter power dissipation heat from each element. This power dissipation is strongly lower-bounded by the signal adopted by present communication standards, which forces an energy efficiency minimum. This also causes a difficult thermal problem in the array implementation. Achieving an energy efficient massive MIMO implementation is shown to require the use of strongly nonlinear transmitters, needing signal designs that are compatible with this nonlinearity while maintaining bandwidth efficiency.

A two-stage D-band differential cascode power amplifier is presented, integrated using the IHP 0.13 pm SiGe BiCMOS technology. A compact layout of the cascode stage is proposed to minimize the parasitics contributing to potential instability, achieving 13 dB of gain/stage while operating at 46% of f T , using stage peaking. The PA is analyzed using Booth chart techniques showing that the amplifier can operate with greatly reduced supply and bias while maintaining linearity dynamic range. The fabricated prototype achieves P1dB, Psat, and IIP2 of 0.2 dBm, 6.2 dBm, and 25.5 dBm respectively with 26 dB overall gain. The AM-PM conversion of the PA is experimentally characterized showing phase fluctuations lower than ±1.5° in a 10 GHz bandwidth. Furthermore, it exhibits 1 dB deviation of power gain over 40 dB input power range with 0.4 V supply variation.

Achieving energy-efficient "Green" operations within data centers used for applications such as Cloud Computing requires matching the power consumed to the data processed at each server in real-time. Beyond having a high efficiency in the server power supplies, it is vitally important to only draw power when the server is actually processing data. To operate at maximum energy efficiency, in the times when a server is idle it needs to draw no power for the server farm. By matching power draw to actual data processing activity at logic speeds, the average energy draw of the server farm drops by 50% or more with no reduction in throughput. Drawing on technology developed for efficient radio transmitters, an agile power supply, able to provide tight voltage regulation and still transition between power-off and power-on (or the other way) in nanoseconds without transition overshoot is described. With this nanosecond agility, this also solves the objective for elastic computing. Additionally, the supply pin pairing required by this energy management method provides benefits toward reducing electromagnetic interference (EMI). Proportional reduction in processor operating temperature improves reliability, along with reducing facility cooling loads.

There is a long history of trading off linearity for better energy efficiency in transmitter design. This tradeoff is fundamental, and is a direct consequence of Ohm's law: if one specification is extremely important, then optimizing that one always forces a minimum of the other. Beginning with representative transistor characteristic curves, the specific physics of this tradeoff between linearity and efficiency are illustrated. Accessing the available high-efficiency options are described, from high-voltage linear design to the ultimate maximum available efficiency which requires strongly nonlinear operation. Use of dynamic power supply operation as a middle ground is explored, requiring a new characterization of power amplifiers as three-port circuits. From this new characterization, multiple power amplifier (PA) operating modes are identified. Properties of these PA operating modes guide the designs which use these modes. Matching available transistor types to these operating modes leads to establishing design preferences based on the performance specifications.

This survey examines the bandwidth efficiency of wireless signals, beginning with a review of the Shannon capacity and of the important differences between signal to noise ratio (snr) and individual bit energy to noise density (IBEND = Eb/N0). Recognizing that communication systems are successful only if they are profitable, value propositions that have been realized along the evolution of adopted wireless signals are reviewed. As the market now desires gigabit data rates while mobile, physical principles are used to extend this historical trend to evaluate options for implementation and their inherent costs so that coming value propositions can be fairly evaluated.

With the evolution of cellular wireless systems and services, the on-air signals themselves are also undergoing very significant transformations. This paper provides a survey of the active and coming-soon signal types adopted for cellular wireless systems around the world. Focus is on modulation schemes, along with various measures used to characterize the signals before and after power amplification. Cost-benefit tradeoff information is introduced to provide perspective on this signal evolution.

In binary FSK modulation schemes the transition between the two frequencies f(0) representing a ZERO and f(1) representing a ONE is governed by the shape of the input pulse as it makes its transition from a ZERO to a ONE. Through the use of an optimum filter, it is possible to shape the pulse in such a way as to maximize the speed of transition from f(0) to f(1) while maintaining signal energy outside of the channel to specified small levels. In this paper, we describe a procedure for finding optimum filters for several important types of FSK modulation.

A general FSK spectrum simulator is written in MATLAB(R) to evaluate the optimum filtering for FSK modulation systems in near-far interference limited channels. In addition, only constant envelope (CE) signals are considered because of their demonstrated energy efficiency. The Fourier Uncertainty Relation is used to evaluate the relative performance of partial response filtering to M-ary signaling. Simulation results are compared with hardware measurement. A similar program is in process for CE-PSK modulation.

Near-far interference is shown to be a potentially serious problem for low cost digital wireless communications systems. Low hardware cost implementations of both binary frequency shift keying (BFSK) and binary phase shift keying (BPSK) transmitters are evaluated. The near-far interference performances of these two modulations are compared. Path loss is the only propagation effect considered. Two significant results of these experiments are: (1) unfiltered BPSK is far more likely to cause near-far interference than unfiltered BFSK, and (2) to avoid near-far interference, modulations must be used that constrain the occupied spectrum within -70 dB to -100 dB relative (dBr) to the signal spectral peak.

- A general FSK spectrum simulator is written in MATLABED to evaluate the optimum filtering for FSK modulation systems in near-far interference limited ChaMelS. In addition, only constant envelope (CE) signals are considered because of their demonstrated energy effciency. The Fourier Uncertainty Relation is used to evaluate the relative performance of partial response filtering to M-ary signaling. Simulation results are compared with hardware measurement. A similar program is in process for CEPSK modulation.

A new model for path loss in radio systems is proposed, which combines the individual segment approach based on varying propagation constants into a single closed form expression. Before this work, the propagation models used examined each propagation segment individually. This closed form expression is more useful than prior models in performing near-far system analyses. This new model is based on concepts drawn from frequency response analysis used in the circuits and control communities. After showing that the general form of a frequency response fits the data shape of a path loss propagation study, the mathematics are refined to effect a close match with measured data. The applicability of this model to signal strengths that increases with distance, such as emerging from a shadow, is demonstrated.