Approximate Computing with Unreliable Dynamic Memories

Embedded memories account for a large fraction of the overall silicon area and power consumption in modern SoC(s). While embedded memories are typically realized with SRAM, alternative solutions, such as embedded dynamic memories (eDRAM), can provide higher density and/or reduced power consumption. One major challenge that impedes the widespread adoption of eDRAM is that they require frequent refreshes potentially reducing the availability of the memory in periods of high activity and also consuming significant amount of power due to such frequent refreshes. Reducing the refresh rate while on one hand can reduce the power overhead, if not performed in a timely manner, can cause some cells to lose their content potentially resulting in memory errors. In this paper, we consider extending the refresh period of gain-cell based dynamic memories beyond the worst-case point of failure, assuming that the resulting errors can be tolerated when the use-cases are in the domain of inherently error-resilient applications. For example, we observe that for various data mining applications, a large number of memory failures can be accepted with tolerable imprecision in output quality. In particular, our results indicate that by allowing as many as 177 errors in a 16 kB memory, the maximum loss in output quality is 11%. We use this failure limit to study the impact of relaxing reliability constraints on memory availability and retention power for different technologies.

Performance assessment of cascade control loopswith non-Gaussian disturbances using entropy information

Cascade control is one of the routinely used control strategies in industrial processes because it can dramatically improve the performance of single-loop control, reducing both the maximum deviation and the integral error of the disturbance response. Currently, many control performance assessment methods of cascade control loops are developed based on the assumption that all the disturbances are subject to Gaussian distribution. However, in the practical condition, several disturbance sources occur in the manipulated variable or the upstream exhibits nonlinear behaviors. In this paper, a general and effective index of the performance assessment of the cascade control system subjected to the unknown disturbance distribution is proposed. Like the minimum variance control (MVC) design, the output variances of the primary and the secondary loops are decomposed into a cascade-invariant and a cascade-dependent term, but the estimated ARMA model for the cascade control loop based on the minimum entropy, instead of the minimum mean squares error, is developed for non-Gaussian disturbances. Unlike the MVC index, an innovative control performance index is given based on the information theory and the minimum entropy criterion. The index is informative and in agreement with the expected control knowledge. To elucidate wide applicability and effectiveness of the minimum entropy cascade control index, a simulation problem and a cascade control case of an oil refinery are applied. The comparison with MVC based cascade control is also included.

Identification and output tracking control of Hammerstein systems through wireless networks

his paper investigates the identification and output tracking control of a class of Hammerstein systems through a wireless network within an integrated framework and the statistic characteristics of the wireless network are modelled using the inverse Gaussian cumulative distribution function. In the proposed framework, a new networked identification algorithm is proposed to compensate for the influence of the wireless network delays so as to acquire the more precise Hammerstein system model. Then, the identified model together with the model-based approach is used to design an output tracking controller. Mean square stability conditions are given using linear matrix inequalities (LMIs) and the optimal controller gains can be obtained by solving the corresponding optimization problem expressed using LMIs. Illustrative numerical simulation examples are given to demonstrate the effectiveness of our proposed method.

Impact of Residual Transmit RF Impairments on Training-Based MIMO Systems

Radio-frequency (RF) impairments, which intimately exist in wireless communication systems, can severely limit the performance of multiple-input-multiple-output (MIMO) systems. Although we can resort to compensation schemes to mitigate some of these impairments, a certain amount of residual impairments always persists. In this paper, we consider a training-based point-to-point MIMO system with residual transmit RF impairments (RTRI) using spatial multiplexing transmission. Specifically, we derive a new linear channel estimator for the proposed model, and show that RTRI create an estimation error floor in the high signal-to-noise ratio (SNR) regime. Moreover, we derive closed-form expressions for the signal-to-noise-plus-interference ratio (SINR) distributions, along with analytical expressions for the ergodic achievable rates of zero-forcing, maximum ratio combining, and minimum mean-squared error receivers, respectively. In addition, we optimize the ergodic achievable rates with respect to the training sequence length and demonstrate that finite dimensional systems with RTRI generally require more training at high SNRs than those with ideal hardware. Finally, we extend our analysis to large-scale MIMO configurations, and derive deterministic equivalents of the ergodic achievable rates. It is shown that, by deploying large receive antenna arrays, the extra training requirements due to RTRI can be eliminated. In fact, with a sufficiently large number of receive antennas, systems with RTRI may even need less training than systems with ideal hardware.

Towards generalized Doherty power amplifier design for multiband operation

This paper presents a new variant of broadband Doherty power amplifier that employs a novel output combiner. A new parameter ∝ is introduced to permit a generalized analysis of the recently reported Parallel Doherty power amplifier (PDPA),and hence offer design flexibility. The circuit prototype of the new DPA fabricated using GaN devices exhibits maximum drain efficiency of 85% at 43-dBm peak power and 63% at 6-dB backoff power (BOP). Measured drain efficiency of >60% at peak power across 500-MHz frequency range and >50% at 6-dB BOP across 480-MHz frequency range were achieved, confirming the  theoretical wideband characteristics of the new DPA.

Sum-Rate and Power Scaling of Massive MIMO Systems with Channel Aging

This paper investigates the achievable sum-rate of massive multiple-input multiple-output (MIMO) systems in the presence of channel aging. For the uplink, by assuming that the base station (BS) deploys maximum ratio combining (MRC) or zero-forcing (ZF) receivers, we present tight closed-form lower bounds on the achievable sum-rate for both receivers with aged channel state information (CSI). In addition, the benefit of implementing channel prediction methods on the sum-rate is examined, and closed-form sum rate lower bounds are derived. Moreover, the impact of channel aging and channel prediction on the power scaling law is characterized. Extension to the downlink scenario and multi-cell scenario are also considered. It is found that, for a system with/without channel prediction, the transmit power of each user can be scaled down at most by 1= p M (where M is the number of BS antennas), which indicates that aged CSI does not degrade the power scaling law, and channel prediction does not enhance the power scaling law; instead, these phenomena affect the achievable sum-rate by degrading or enhancing the effective signal to interference and noise ratio, respectively.

Multi-pair two-way AF relaying with massive arrays and imperfect CSI

We consider a multi-pair two-way amplify-and-forward relaying system with a massive antenna array at the relay and estimated channel state information, assuming maximum-ratio combining/transmission processing. Closed-form approximations of the sum spectral effi- ciency are developed and simple analytical power scaling laws are presented, which reveal a fundamental trade-off between the transmit powers of each user/the relay and of each pilot symbol. Finally, the optimal power allocation problem is studied.

Computing probable maximum loss in catastrophe reinsurance portfolios on multi-core and many-core architectures

In the reinsurance market, the risks natural catastrophes pose to portfolios of properties must be quantified, so that they can be priced, and insurance offered. The analysis of such risks at a portfolio level requires a simulation of up to 800 000 trials with an average of 1000 catastrophic events per trial. This is sufficient to capture risk for a global multi-peril reinsurance portfolio covering a range of perils including earthquake, hurricane, tornado, hail, severe thunderstorm, wind storm, storm surge and riverine flooding, and wildfire. Such simulations are both computation and data intensive, making the application of high-performance computing techniques desirable.

In this paper, we explore the design and implementation of portfolio risk analysis on both multi-core and many-core computing platforms. Given a portfolio of property catastrophe insurance treaties, key risk measures, such as probable maximum loss, are computed by taking both primary and secondary uncertainties into account. Primary uncertainty is associated with whether or not an event occurs in a simulated year, while secondary uncertainty captures the uncertainty in the level of loss due to the use of simplified physical models and limitations in the available data. A combination of fast lookup structures, multi-threading and careful hand tuning of numerical operations is required to achieve good performance. Experimental results are reported for multi-core processors and systems using NVIDIA graphics processing unit and Intel Phi many-core accelerators.

Factors influencing the output of rural cataract surgical facilities in China: the SHARP study

Purpose: To identify factors associated prospectively with increased cataract surgical rate (CSR) in rural Chinese hospitals.

Methods: Annual cataract surgical output was obtained at baseline and 24 months later from operating room records at 42 rural, county-level hospitals. Total local CSR (cases/million population/y), and proportion of CSR from hospital and local competitors were calculated from government records. Hospital administrators completed questionnaires providing demographic and professional information, and annual clinic and outreach screening volume. Independent cataract surgeons provided clinical information and videotapes of cases for grading by two masked experts using the Ophthalmology Surgical Competency Assessment Rubric (OSCAR). Uncorrected vision was recorded for 10 consecutive cataract cases at each facility, and 10 randomly-identified patients completed hospital satisfaction questionnaires. Total value of international nongovernmental development organization (INGDO) investment in the previous three years and demographic information on hospital catchment areas were obtained. Main outcome was 2-year percentage change in hospital CSR.

Results: Among the 42 hospitals (median catchment population 530,000, median hospital CSR 643), 78.6% (33/42) were receiving INGDO support. Median change in hospital CSR (interquartile range) was 33.3% (-6.25%, 72.3%). Predictors of greater increase in CSR included higher INGDO investment (P = 0.02, simple model), reducing patient dissatisfaction (P = 0.03, simple model), and more outreach patient screening (P = 0.002, simple and multiple model).

Conclusions: Outreach cataract screening was the strongest predictor of increased surgical output. Government and INGDO investment in screening may be most likely to enhance output of county hospitals, a major goal of China's Blindness Prevention Plan.

Characterisation of a two-dimensional liquid-filled ion chamber detector array using flattened and unflattened beams for small fields, small MUs and high dose-rates

In this study, the PTW 1000SRS array with Octavius 4D phantom was characterised for FF and FFF beams. MU linearity, field size, dose rate, dose per pulse (DPP) response and dynamic conformal arc treatment accuracy of the 1000SRS array were assessed for 6MV, 6FFF and 10FFF beams using a Varian TrueBeam STx linac. The measurements were compared with a pinpoint IC, microdiamond IC and EBT3 Gafchromic film. Measured dose profiles and FWHMs were compared with film measurements. Verification of FFF volumetric modulated arc therapy (VMAT) clinical plans were assessed using gamma analysis with 3%/3 mm and 2%/2 mm tolerances (10% threshold). To assess the effect of cross calibration dose rate, clinical plans with different dose rates were delivered and analysed. Output factors agreed with film measurements to within 4.5% for fields between 0.5 and 1 cm and within 2.7% for field sizes between 1.5 and 10 cm and were highly correlated with the microdiamond IC detector. Field sizes measured with the 1000SRS array were within 0.5 mm of film measurements. A drop in response of up to 1.8%, 2.4% and 5.2% for 6MV, 6FFF and 10FFF beams respectively was observed with increasing nominal dose rate. With an increase in DPP, a drop of up to 1.7%, 2.4% and 4.2% was observed in 6MV, 6FFF and 10FFF respectively. The differences in dose following dynamic conformal arc deliveries were less than 1% (all energies) from calculated. Delivered VMAT plans showed an average pass percentage of 99.5(±0.8)% and 98.4(±3.4)% with 2%/2 mm criteria for 6FFF and 10FFF respectively. A drop to 97.7(±2.2)% and 88.4(±9.6)% were observed for 6FFF and 10FFF respectively when plans were delivered at the minimum dose rate and calibrated at the maximum dose rate. Calibration using a beam with the average dose rate of the plan may be an efficient method to overcome the dose rate effects observed by the 1000SRS array.

IQ Imbalance in Multiuser Systems: Channel Estimation and Compensation

In this paper, we consider the uplink of a single-cell multi-user single-input multiple-output (MU-SIMO) system with in-phase and quadrature-phase imbalance (IQI). Particularly, we investigate the effect of receive (RX) IQI on the performance of MU-SIMO systems with large antenna arrays employing maximum-ratio combining (MRC) receivers. In order to study how IQI affects channel estimation, we derive a new channel estimator for the IQI-impaired model and show that the higher the value of signal-to-noise ratio (SNR) the higher the impact of IQI on the spectral efficiency (SE). Moreover, a novel pilot-based joint estimator of the augmented MIMO channel matrix and IQI coefficients is described and then, a low-complexity IQI compensation scheme is proposed which is based on the
IQI coefficients’ estimation and it is independent of the channel gain. The performance of the proposed compensation scheme is analytically evaluated by deriving a tractable approximation of the ergodic SE assuming transmission over Rayleigh fading channels with large-scale fading. Furthermore, we investigate how many MSs should be scheduled in massive multiple-input multiple-output (MIMO) systems with IQI and show that the highest SE loss occurs at the optimal operating point. Finally,
by deriving asymptotic power scaling laws, and proving that the SE loss due to IQI is asymptotically independent of the number of BS antennas, we show that massive MIMO is resilient to the effect of RX IQI.

Convex Quadratic Programming Approach to the Maximum Matching Problem

The problem of determining a maximum matching or whether there exists a perfect matching, is very common in a large variety of applications and as been extensively studied in graph theory. In this paper we start to introduce a characterisation of a family of graphs for which its stability number is determined by convex quadratic programming. The main results connected with the recognition of this family of graphs are also introduced. It follows a necessary and sufficient condition which characterise a graph with a perfect matching and an algorithmic strategy, based on the determination of the stability number of line graphs, by convex quadratic programming, applied to the determination of a perfect matching. A numerical example for the recognition of graphs with a perfect matching is described. Finally, the above algorithmic strategy is extended to the determination of a maximum matching of an arbitrary graph and some related results are presented.

Polarization effects in fiber-optic communication systems

In this thesis we perform a detailed analysis of the state of polarization (SOP) of light scattering process using a concatenation of ber-coil based polarization controllers (PCs). We propose a polarization-mode dispersion (PMD) emulator, built through the concatenation of bercoil based PCs and polarization-maintaining bers (PMFs), capable of generate accurate rst- and second-order PMD statistics. We analyze the co-propagation of two optical waves inside a highbirefringence ber. The evolution along the ber of the relative SOP between the two signals is modeled by the de nition of the degree of co-polarization parameter. We validate the model for the degree of co-polarization experimentally, exploring the polarization dependence of the four-wave mixing e ect into a ber with high birefringence. We also study the interaction between signal and noise mediated by Kerr e ect in optical bers. A model accurately describing ampli ed spontaneous emission noise in systems with distributed Raman gain is derived. We show that the noise statistics depends on the propagation distance and on the signal power, and that for distances longer than 120 km and signal powers higher than 6 mW it deviates signi catively from the Gaussian distribution. We explore the all-optical polarization control process based on the stimulated Raman scattering e ect. Mapping parameters like the degree of polarization (DOP), we show that the preferred ampli cation of one particular polarization component of the signal allows a polarization pulling over a wavelength range of 60 nm. The e ciency of the process is higher close to the maximum Raman gain wavelength, where the DOP is roughly constant for a wavelength range of 15 nm. Finally, we study the polarization control in quantum key distribution (QKD) systems with polarization encoding. A model for the quantum bit error rate estimation in QKD systems with time-division multiplexing and wavelength-division multiplexing based polarization control schemes is derived.