As communication and computation systems become more complex and target at higher performance, the fundamental limits of nature can be expected to constrain their development and optimization. This calls for intelligent use of basic resources, that is, materials, energy, information, time, frequency, and space. We present a multidisciplinary and historical review on the body of knowledge that can be applied in researching such intelligent and resource-efficient systems. We review general system theory, decision theory, control theory, computer science, and communication theory. While multidisciplinarity has been recognized important, there are no earlier reviews covering all these five disciplines. Based on the review, we build a chronology of intelligent systems and identify connections between the disciplines. Optimization, decision-making, open- and closed-loop control, hierarchy, and degree of centralization turn out as recurring themes in these disciplines, which have converged to similar solutions that are based on remote control, automation, autonomy, and self-organization. We use future wireless networks as an example to illustrate the open questions and how they can be addressed by applying multidisciplinary knowledge. This paper can help researchers to use knowledge outside their own field and avoid repeating the work done already. The resulting consolidated view can speed up research and is especially important when the fundamental limits of nature are approached and new insights are required to overcome the challenges. The general, long-standing problem to be tackled is multiobjective optimization with autonomous and distributed decision-making in an uncertain, dynamic, and nonlinear environment where the objectives are mutually conflicting.

The threat of quantum-computer-assisted cryptanalysis is forcing the security community to develop new types of security protocols. These solutions must be secure against classical and post-quantum cryptanalysis techniques as well as feasible for all kinds of devices, including energy-restricted Internet of Things (IoT) devices. The quantum immunity can be implemented in the cryptographic layer, e.g., by using recent lattice-based key exchange algorithms NewHope or Frodo, or in the physical layer of wireless communication, by utilizing eavesdropping-resistant secrecy coding techniques. In this study, we explore and compare the feasibility and energy efficiency of selected cryptographic layer and physical layer approaches by applying an evaluation approach that is based on simulation and modeling. In particular, we consider NewHope and Frodo key exchange algorithms as well as novel physical layer secrecy coding approach that is based on polar codes. The results reveal that our proposed physical layer implementation is very competitive with respect to the cryptographic solutions, particularly in short-range wireless communication. We also observed that the total energy consumption is unequally divided between transmitting and receiving devices in all the studied approaches. This may be an advantage when designing security architectures for energy-restricted devices.

Noncoherent multilevel systems have been typically analyzed with spectrally inefficient orthogonal modulation methods. In this paper, we present a novel error probability analysis of energy detected (ED) signals with spectrally efficient multilevel pulse amplitude modulation (PAM). One of the main challenges in the analysis of ED systems with the multilevel PAM is to find analytical methods to evaluate and optimize the performance of ED systems with respect to arbitrary system parameters, e.g., decision thresholds for chi-squared-distributed decision variables, integration time, bandwidth, and number of modulation levels. We propose analytically tractable methods that can be efficiently used to design spectrally efficient ED systems for short-range, wideband, and high data rate wireless communications in lognormal multipath fading channels with uncorrelated diversity paths.

This paper studies the analytical relationship between the average transmitted and received energies under several adaptive transmitter power control methods, including water filling, truncated power inversion, and downlink beamforming. The study is applicable to many fading channel scenarios, including frequency-nonselective, frequency-selective, and multiple-input-multiple-output (MIMO) channels. Both the average transmitted and received energies are commonly used in performance comparisons, and the selection depends on what one wants to investigate. The transmitted energy is known to be the basic system resource. In the case of adaptive transmission, the average transmitted energy should, in general, be used instead of the average received energy. The use of transmitted energy leads to the normalization problem of the channel. The ratio of received energy to transmitted energy is the energy gain of the channel. All physical systems follow an energy-conservation law, which implies that the energy gain of the channel is less than or equal to 1. The major approaches for normalization include the setting of either the average energy gain or the peak energy gain to unity. In the normalization, the average energy gain is defined for a signal whose energy is uniformly distributed across the frequency and spatial dimensions. The peak energy gain of many mathematical fading models is not bounded, and those models cannot be normalized by the peak energy gain. We show that the proper normalization of the mathematical model and the selection of the correct performance measure are of critical importance in comparative performance analysis of adaptive transmission systems.

In this paper we theoretically evaluate the sensitivity resulted from a decision threshold mismatch with multilevel and energy detected pulse amplitude modulated (ED-PAM) signals. We quantify the relation between the threshold offset and the degradation in the required signal-to-noise ratio to preserve a desired symbol error rate. Based on the analysis we conclude some useful guidelines with respect to various system parameters. One of the main problems in an ED-PAM with arbitrary number of modulation levels, integration time, and bandwidth is to find a simple way to calculate the optimal threshold values. We rise to this challenge by proposing a new multilevel threshold calculation method which is attractive to use from both performance and complexity point of views for a wide range of system parameters. The results can be used, e.g., in an impulse radio for high data rate short-range wireless communications.

Entropy in workplaces is situated amidst workers and their work. In this paper, findings are reported from a study encompassing psychomotor work by three types of workers: human, cyborg and robot; together with three aspects of psychomotor work: setting, composition and uncertainty. The Principle of Least Psychomotor Action (PLPA) is introduced and modelled in terms of situated entropy. PLPA is founded upon the Principle of Least Action. Situated entropy modelling of PLPA is informed by theoretical studies concerned with connections between information theory and thermodynamics. Four contributions are provided in this paper. First, the situated entropy of PLPA is modelled in terms of positioning, performing and perfecting psychomotor skills. Second, with regard to workers, PLPA is related to the state-of-the-art in human, cyborg and robot psychomotor skills. Third, with regard to work, situated entropy is related to engineering of work settings, work composition and work uncertainty. Fourth, PLPA and modelling situated entropy are related to debate about the future of work. Overall, modelling situated entropy is introduced as a means of objectively modelling relative potential of humans, cyborgs, and robots to carry out work with least action. This can introduce greater objectivity into debates about the future of work.