Fault Tolerance and Noise Immunity in Freespace Diffractive Optical Neural Networks

Free-space diffractive optical networks are a class of trainable optical media that are currently being explored as a novel hardware platform for neural engines. The training phase of such systems is usually performed in a computer and the learned weights are then transferred onto optical hardware ("ex-situ training"). Although this process of weight transfer has many practical advantages, it is often accompanied by performance degrading faults in the fabricated hardware. Being analog systems, these engines are also subject to performance degradation due to noises in the inputs and during optoelectronic conversion. Considering diffractive optical networks (DON) trained for image classification tasks on standard datasets, we numerically study the performance degradation arising out of weight faults and injected noises and methods to ameliorate these effects. Training regimens based on intentional fault and noise injection during the training phase are only found marginally successful at imparting fault tolerance or noise immunity. We propose an alternative training regimen using gradient based regularization terms in the training objective that are found to impart some degree of fault tolerance and noise immunity in comparison to injection based training regimen.

Study of Resistance to Disturbances of the Main Types of Communication Systems on Board Military Ships Used during Interception or Search and Rescue Missions

In addition to combat missions, military ships often participate in search-and-rescue missions or interception of ships with refugees or migrants. The communication systems and modes of work that may be used during these missions will be in accordance with the International Convention for the Safety of Life at Sea (SOLAS). This paper aims to demonstrate by theoretical methods (analytical and numerical) the noise stability of communication systems using analog modulation in high-noise conditions, characteristic of the marine environment. The stability of analog systems employing amplitude (AM), frequency (FM), and phase (PM) modulations is investigated. The analyzed systems are currently under use in distress maritime communications.

Behavioral responses to a cyber attack in a hospital environment

Technical and organizational steps are necessary to mitigate cyber threats and reduce risks. Human behavior is the last line of defense for many hospitals and is considered as equally important as technical security. Medical staff must be properly trained to perform such procedures. This paper presents the first qualitative, interdisciplinary research on how members of an intermediate care unit react to a cyberattack against their patient monitoring equipment. We conducted a simulation in a hospital training environment with 20 intensive care nurses. By the end of the experiment, 12 of the 20 participants realized the monitors’ incorrect behavior. We present a qualitative behavior analysis of high performing participants (HPP) and low performing participants (LPP). The HPP showed fewer signs of stress, were easier on their colleagues, and used analog systems more often than the LPP. With 40% of our participants not recognizing the attack, we see room for improvements through the use of proper tools and provision of adequate training to prepare staff for potential attacks in the future.

Order-of-magnitude differences in computational performance of analog Ising machines induced by the choice of nonlinearity

Ising machines based on nonlinear analog systems are a promising method to accelerate computation of NP-hard optimization problems. Yet, their analog nature is also causing amplitude inhomogeneity which can deteriorate the ability to find optimal solutions. Here, we investigate how the system’s nonlinear transfer function can mitigate amplitude inhomogeneity and improve computational performance. By simulating Ising machines with polynomial, periodic, sigmoid and clipped transfer functions and benchmarking them with MaxCut optimization problems, we find the choice of transfer function to have a significant influence on the calculation time and solution quality. For periodic, sigmoid and clipped transfer functions, we report order-of-magnitude improvements in the time-to-solution compared to conventional polynomial models, which we link to the suppression of amplitude inhomogeneity induced by saturation of the transfer function. This provides insights into the suitability of nonlinear systems for building Ising machines and presents an efficient way for overcoming performance limitations.

Characterization of Nanocrystalline Cores for EMI Suppression in Cables

Electromagnetic interferences (EMI) can cause different kinds of problems in digital and analog systems, leading to malfunctions, system reboots, or even permanent damage to the system if this is not adequately designed or protected. Nowadays, most electronic products are connected to the main power network or are designed to be interconnected with others through cables. These cable interconnections are becoming more difficult due to the rigid restrictions related to the accomplishment of electromagnetic compatibility (EMC) compliance. When the cables of a system represent an EMI source, it cannot pass the conducted or radiated emissions test. A widely used technique to reduce these problems is applying an EMI suppressor such as a sleeve core. This EMI suppressor provides selective attenuation of undesired interference components that the designer may wish to suppress, and it does not significantly affect the intended signal. This contribution focuses on analyzing different nanocrystalline (NC) EMI suppressors’ performance intended for attenuating interferences in cables. Some NC novel samples are characterized and compare to MnZn and NiZn cores to determine this novel material’s effectiveness compared to the conventional ceramic solutions by analyzing samples with different dimensions.

Digital On-Chip Calibration of Analog Systems towards Enhanced Reliability

This chapter deals with digital method of calibration for analog integrated circuits as a means of extending its lifetime and reliability, which consequently affects the reliability the analog electronic system as a whole. The proposed method can compensate for drift in circuit’s electrical parameters, which occurs either in a long term due to aging and electrical stress or it is rather more acute, being caused by process, voltage and temperature variations. The chapter reveals the implementation of ultra-low voltage on-chip system of digitally calibrated variable-gain amplifier (VGA), fabricated in CMOS 130 nm technology. It operates reliably under supply voltage of 600 mV with 10% variation, in temperature range from − 20 ° C to 85 ° C . Simulations suggest that the system will preserve its parameters for at least 10 years of operation. Experimental verification over 10 packaged integrated circuit (IC) samples shows the input offset voltage of VGA is suppressed in range of 13 μV to 167 μV . With calibration the VGA closely meets its nominally designed essential specifications as voltage gain or bandwidth. Digital calibration is comprehensively compared to its widely used alternative, Chopper stabilization through its implementation for the same VGA.

Prospective study of dynamic whole-body 68Ga-DOTATOC-PET/CT acquisition in patients with well-differentiated neuroendocrine tumors

To present the feasibility of a dynamic whole-body (DWB) 68Ga-DOTATOC-PET/CT acquisition in patients with well-differentiated neuroendocrine tumors (WD-NETs). Sixty-one patients who underwent a DWB 68Ga-DOTATOC-PET/CT for a histologically proven/highly suspected WD-NET were prospectively included. The acquisition consisted in single-bed dynamic acquisition centered on the heart, followed by the DWB and static acquisitions. For liver, spleen and tumor (1–5/patient), Ki values (in ml/min/100 ml) were calculated according to Patlak's analysis and tumor-to-liver (TLR-Ki) and tumor-to-spleen ratios (TSR-Ki) were recorded. Ki-based parameters were compared to static parameters (SUVmax/SUVmean, TLR/TSRmean, according to liver/spleen SUVmean), in the whole-cohort and according to the PET system (analog/digital). A correlation analysis between SUVmean/Ki was performed using linear and non-linear regressions. Ki-liver was not influenced by the PET system used, unlike SUVmax/SUVmean. The regression analysis showed a non-linear relation between Ki/SUVmean (R2 = 0.55,0.68 and 0.71 for liver, spleen and tumor uptake, respectively) and a linear relation between TLRmean/TLR-Ki (R2 = 0.75). These results were not affected by the PET system, on the contrary of the relation between TSRmean/TSR-Ki (R2 = 0.94 and 0.73 using linear and non-linear regressions in digital and analog systems, respectively). Our study is the first showing the feasibility of a DWB 68Ga-DOTATOC-PET/CT acquisition in WD-NETs.

Using noise to augment synchronization among oscillators

Noise is expected to play an important role in the dynamics of analog systems such as coupled oscillators which have recently been explored as a hardware platform for application in computing. In this work, we experimentally investigate the effect of noise on the synchronization of relaxation oscillators and their computational properties. Specifically, in contrast to its typically expected adverse effect, we first demonstrate that a common white noise input induces frequency locking among uncoupled oscillators. Experiments show that the minimum noise voltage required to induce frequency locking increases linearly with the amplitude of the oscillator output whereas it decreases with increasing number of oscillators. Further, our work reveals that in a coupled system of oscillators—relevant to solving computational problems such as graph coloring, the injection of white noise helps reduce the minimum required capacitive coupling strength. With the injection of noise, the coupled system demonstrates frequency locking along with the desired phase-based computational properties at 5 × lower coupling strength than that required when no external noise is introduced. Consequently, this can reduce the footprint of the coupling element and the corresponding area-intensive coupling architecture. Our work shows that noise can be utilized as an effective knob to optimize the implementation of coupled oscillator-based computing platforms.

High throughput error correction in information reconciliation for semiconductor superlattice secure key distribution

Semiconductor superlattice secure key distribution (SSL-SKD) has been experimentally demonstrated to be a novel scheme to generate and agree on the identical key in unconditional security just by public channel. The error correction in the information reconciliation procedure is introduced to eliminate the inevitable differences of analog systems in SSL-SKD. Nevertheless, the error correction has been proved to be the performance bottleneck of information reconciliation for high computational complexity. Hence, it determines the final secure key throughput of SSL-SKD. In this paper, different frequently-used error correction codes, including BCH codes, LDPC codes, and Polar codes, are optimized separately to raise the performance, making them usable in practice. Firstly, we perform multi-threading to support multi-codeword decoding for BCH codes and Polar codes and updated value calculation for LDPC codes. Additionally, we construct lookup tables to reduce redundant calculations, such as logarithmic table and antilogarithmic table for finite field computation. Our experimental results reveal that our proposed optimization methods can significantly promote the efficiency of SSL-SKD, and three error correction codes can reach the throughput of Mbps and provide a minimum secure key rate of 99%.