AdFT: An Adaptive Framework for Fault Tolerance on Large Scale Systems using Application Malleability

Exascale systems of the future are predicted to have mean time between failures (MTBF) of less than one hour. Malleable applications, where the number of processors on which the applications execute can be changed during executions, can make use of their malleability to better tolerate high failure rates. We present AdFT, an adaptive fault tolerance framework for long running malleable applications to maximize application performance in the presence of failures. AdFT framework includes cost models for evaluating the benefits of various fault tolerance actions including checkpointing, live-migration and rescheduling, and runtime decisions for dynamically selecting the fault tolerance actions at different points of application execution to maximize performance. Simulations with real and synthetic failure traces show that our approach outperforms existing fault tolerance mechanisms for malleable applications yielding up to 23% improvement in application performance, and is effective even for petascale systems and beyond.

Fault Tolerance on Large Scale Systems using Adaptive Process Replication

Exascale systems of the future are predicted to have mean time between failures (MTBF) of less than one hour. At such low MTBFs, employing periodic checkpointing alone will result in low efficiency because of the high number of application failures resulting in large amount of lost work due to rollbacks. In such scenarios, it is highly necessary to have proactive fault tolerance mechanisms that can help avoid significant number of failures. In this work, we have developed a mechanism for proactive fault tolerance using partial replication of a set of application processes. Our fault tolerance framework adaptively changes the set of replicated processes periodically based on failure predictions to avoid failures. We have developed an MPI prototype implementation, PAREP-MPI that allows changing the replica set. We have shown that our strategy involving adaptive process replication significantly outperforms existing mechanisms providing up to 20 percent improvement in application efficiency even for exascale systems.

A Novel Monte-Carlo-Sampling-Based Receiver for Large-Scale Uplink Multiuser MIMO Systems

In this paper, we propose low-complexity algorithms based on Monte Carlo sampling for signal detection and channel estimation on the uplink in large-scale multiuser multiple-input-multiple-output (MIMO) systems with tens to hundreds of antennas at the base station (BS) and a similar number of uplink users. A BS receiver that employs a novel mixed sampling technique (which makes a probabilistic choice between Gibbs sampling and random uniform sampling in each coordinate update) for detection and a Gibbs-sampling-based method for channel estimation is proposed. The algorithm proposed for detection alleviates the stalling problem encountered at high signal-to-noise ratios (SNRs) in conventional Gibbs-sampling-based detection and achieves near-optimal performance in large systems with M-ary quadrature amplitude modulation (M-QAM). A novel ingredient in the detection algorithm that is responsible for achieving near-optimal performance at low complexity is the joint use of a mixed Gibbs sampling (MGS) strategy coupled with a multiple restart (MR) strategy with an efficient restart criterion. Near-optimal detection performance is demonstrated for a large number of BS antennas and users (e. g., 64 and 128 BS antennas and users). The proposed Gibbs-sampling-based channel estimation algorithm refines an initial estimate of the channel obtained during the pilot phase through iterations with the proposed MGS-based detection during the data phase. In time-division duplex systems where channel reciprocity holds, these channel estimates can be used for multiuser MIMO precoding on the downlink. The proposed receiver is shown to achieve good performance and scale well for large dimensions.

A novel MCMC algorithm for near-optimal detection in large-scale uplink mulituser MIMO systems

In this paper, we propose a low-complexity algorithm based on Markov chain Monte Carlo (MCMC) technique for signal detection on the uplink in large scale multiuser multiple input multiple output (MIMO) systems with tens to hundreds of antennas at the base station (BS) and similar number of uplink users. The algorithm employs a randomized sampling method (which makes a probabilistic choice between Gibbs sampling and random sampling in each iteration) for detection. The proposed algorithm alleviates the stalling problem encountered at high SNRs in conventional MCMC algorithm and achieves near-optimal performance in large systems with M-QAM. A novel ingredient in the algorithm that is responsible for achieving near-optimal performance at low complexities is the joint use of a randomized MCMC (R-MCMC) strategy coupled with a multiple restart strategy with an efficient restart criterion. Near-optimal detection performance is demonstrated for large number of BS antennas and users (e.g., 64, 128, 256 BS antennas/users).

Channel Hardening-Exploiting Message Passing (CHEMP) Receiver in Large-Scale MIMO Systems

In this paper, we propose a multiple-input multiple-output (MIMO) receiver algorithm that exploits channel hardening that occurs in large MIMO channels. Channel hardening refers to the phenomenon where the off-diagonal terms of the matrix become increasingly weaker compared to the diagonal terms as the size of the channel gain matrix increases. Specifically, we propose a message passing detection (MPD) algorithm which works with the real-valued matched filtered received vector (whose signal term becomes, where is the transmitted vector), and uses a Gaussian approximation on the off-diagonal terms of the matrix. We also propose a simple estimation scheme which directly obtains an estimate of (instead of an estimate of), which is used as an effective channel estimate in the MPD algorithm. We refer to this receiver as the channel hardening-exploiting message passing (CHEMP) receiver. The proposed CHEMP receiver achieves very good performance in large-scaleMIMO systems (e.g., in systems with 16 to 128 uplink users and 128 base station antennas). For the considered large MIMO settings, the complexity of the proposed MPD algorithm is almost the same as or less than that of the minimum mean square error (MMSE) detection. This is because the MPD algorithm does not need a matrix inversion. It also achieves a significantly better performance compared to MMSE and other message passing detection algorithms using MMSE estimate of. Further, we design optimized irregular low density parity check (LDPC) codes specific to the considered large MIMO channel and the CHEMP receiver through EXIT chart matching. The LDPC codes thus obtained achieve improved coded bit error rate performance compared to off-the-shelf irregular LDPC codes.

Generalized Spatial Modulation in Large-Scale Multiuser MIMO Systems

Generalized spatial modulation (GSM) uses n(t) transmit antenna elements but fewer transmit radio frequency (RF) chains, n(rf). Spatial modulation (SM) and spatial multiplexing are special cases of GSM with n(rf) = 1 and n(rf) = n(t), respectively. In GSM, in addition to conveying information bits through n(rf) conventional modulation symbols (for example, QAM), the indices of the n(rf) active transmit antennas also convey information bits. In this paper, we investigate GSM for large-scale multiuser MIMO communications on the uplink. Our contributions in this paper include: 1) an average bit error probability (ABEP) analysis for maximum-likelihood detection in multiuser GSM-MIMO on the uplink, where we derive an upper bound on the ABEP, and 2) low-complexity algorithms for GSM-MIMO signal detection and channel estimation at the base station receiver based on message passing. The analytical upper bounds on the ABEP are found to be tight at moderate to high signal-to-noise ratios (SNR). The proposed receiver algorithms are found to scale very well in complexity while achieving near-optimal performance in large dimensions. Simulation results show that, for the same spectral efficiency, multiuser GSM-MIMO can outperform multiuser SM-MIMO as well as conventional multiuser MIMO, by about 2 to 9 dB at a bit error rate of 10(-3). Such SNR gains in GSM-MIMO compared to SM-MIMO and conventional MIMO can be attributed to the fact that, because of a larger number of spatial index bits, GSM-MIMO can use a lower-order QAM alphabet which is more power efficient.

An FETI-preconditioned conjuerate gradient method for large-scale stochastic finite element problems

In the spectral stochastic finite element method for analyzing an uncertain system. the uncertainty is represented by a set of random variables, and a quantity of Interest such as the system response is considered as a function of these random variables Consequently, the underlying Galerkin projection yields a block system of deterministic equations where the blocks are sparse but coupled. The solution of this algebraic system of equations becomes rapidly challenging when the size of the physical system and/or the level of uncertainty is increased This paper addresses this challenge by presenting a preconditioned conjugate gradient method for such block systems where the preconditioning step is based on the dual-primal finite element tearing and interconnecting method equipped with a Krylov subspace reusage technique for accelerating the iterative solution of systems with multiple and repeated right-hand sides. Preliminary performance results on a Linux Cluster suggest that the proposed Solution method is numerically scalable and demonstrate its potential for making the uncertainty quantification Of realistic systems tractable.

Meridional propagation of large-scale monsoon convective zones

Observational studies indicate that the convective activity of the monsoon systems undergo intraseasonal variations with multi-week time scales. The zone of maximum monsoon convection exhibits substantial transient behavior with successive propagating from the North Indian Ocean to the heated continent. Over South Asia the zone achieves its maximum intensity. These propagations may extend over 3000 km in latitude and perhaps twice the distance in longitude and remain as coherent entities for periods greater than 2-3 weeks. Attempts to explain this phenomena using simple ocean-atmosphere models of the monsoon system had concluded that the interactive ground hydrology so modifies the total heating of the atmosphere that a steady state solution is not possible, thus promoting lateral propagation. That is, the ground hydrology forces the total heating of the atmosphere and the vertical velocity to be slightly out of phase, causing a migration of the convection towards the region of maximum heating. Whereas the lateral scale of the variations produced by the Webster (1983) model were essentially correct, they occurred at twice the frequency of the observed events and were formed near the coastal margin, rather than over the ocean. Webster's (1983) model used to pose the theories was deficient in a number of aspects. Particularly, both the ground moisture content and the thermal inertia of the model were severely underestimated. At the same time, the sea surface temperatures produced by the model between the equator and the model's land-sea boundary were far too cool. Both the atmosphere and the ocean model were modified to include a better hydrological cycle and ocean structure. The convective events produced by the modified model possessed the observed frequency and were generated well south of the coastline. The improved simulation of monsoon variability allowed the hydrological cycle feedback to be generalized. It was found that monsoon variability was constrained to lie within the bounds of a positive gradient of a convective intensity potential (I). The function depends primarily on the surface temperature, the availability of moisture and the stability of the lower atmosphere which varies very slowly on the time scale of months. The oscillations of the monsoon perturb the mean convective intensity potential causing local enhancements of the gradient. These perturbations are caused by the hydrological feedbacks, discussed above, or by the modification of the air-sea fluxes caused by variations of the low level wind during convective events. The final result is the slow northward propagation of convection within an even slower convective regime. The ECMWF analyses show very similar behavior of the convective intensity potential. Although it is considered premature to use the model to conduct simulations of the African monsoon system, the ECMWF analysis indicates similar behavior in the convective intensity potential suggesting, at least, that the same processes control the low frequency structure of the African monsoon. The implications of the hypotheses on numerical weather prediction of monsoon phenomenon are discussed.

A Randomized Algorithm for Large Scale Support Vector Learning

We propose a randomized algorithm for large scale SVM learning which solves the problem by iterating over random subsets of the data. Crucial to the algorithm for scalability is the size of the subsets chosen. In the context of text classification we show that, by using ideas from random projections, a sample size of O(log n) can be used to obtain a solution which is close to the optimal with a high probability. Experiments done on synthetic and real life data sets demonstrate that the algorithm scales up SVM learners, without loss in accuracy. 1

Mining large-scale response networks reveals `topmost activities' in Mycobacterium tuberculosis infection

Mycobacterium tuberculosis owes its high pathogenic potential to its ability to evade host immune responses and thrive inside the macrophage. The outcome of infection is largely determined by the cellular response comprising a multitude of molecular events. The complexity and inter-relatedness in the processes makes it essential to adopt systems approaches to study them. In this work, we construct a comprehensive network of infection-related processes in a human macrophage comprising 1888 proteins and 14,016 interactions. We then compute response networks based on available gene expression profiles corresponding to states of health, disease and drug treatment. We use a novel formulation for mining response networks that has led to identifying highest activities in the cell. Highest activity paths provide mechanistic insights into pathogenesis and response to treatment. The approach used here serves as a generic framework for mining dynamic changes in genome-scale protein interaction networks.

Origin of 1/f noise in graphene produced for large-scale applications in electronics

The authors report a detailed investigation of the flicker noise (1/f noise) in graphene films obtained from chemical vapour deposition (CVD) and chemical reduction of graphene oxide. The authors find that in the case of polycrystalline graphene films grown by CVD, the grain boundaries and other structural defects are the dominant source of noise by acting as charged trap centres resulting in huge increase in noise as compared with that of exfoliated graphene. A study of the kinetics of defects in hydrazine-reduced graphene oxide (RGO) films as a function of the extent of reduction showed that for longer hydrazine treatment time strong localised crystal defects are introduced in RGO, whereas the RGO with shorter hydrazine treatment showed the presence of large number of mobile defects leading to higher noise amplitude.

Large-Scale Multiuser SM-MIMO Versus Massive MIMO

Spatial modulation (SM) is attractive for multiantenna wireless communications. SM uses multiple transmit antenna elements but only one transmit radio frequency (RF) chain. In SM, in addition to the information bits conveyed through conventional modulation symbols (e.g., QAM), the index of the active transmit antenna also conveys information bits. In this paper, we establish that SM has significant signal-to-noise (SNR) advantage over conventional modulation in large-scale multiuser (multiple-input multiple-output) MIMO systems. Our new contribution in this paper addresses the key issue of large-dimension signal processing at the base station (BS) receiver (e.g., signal detection) in large-scale multiuser SM-MIMO systems, where each user is equipped with multiple transmit antennas (e.g., 2 or 4 antennas) but only one transmit RF chain, and the BS is equipped with tens to hundreds of (e.g., 128) receive antennas. Specifically, we propose two novel algorithms for detection of large-scale SM-MIMO signals at the BS; one is based on message passing and the other is based on local search. The proposed algorithms achieve very good performance and scale well. For the same spectral efficiency, multiuser SM-MIMO outperforms conventional multiuser MIMO (recently being referred to as massive MIMO) by several dBs. The SNR advantage of SM-MIMO over massive MIMO can be attributed to: (i) because of the spatial index bits, SM-MIMO can use a lower-order QAM alphabet compared to that in massive MIMO to achieve the same spectral efficiency, and (ii) for the same spectral efficiency and QAM size, massive MIMO will need more spatial streams per user which leads to increased spatial interference.

Hydroclimatic influence of large-scale circulation on the variability of reservoir inflow

In this study, the nature of basin-scale hydroclimatic association for Indian subcontinent is investigated. It is found that, the large-scale circulation information from Indian Ocean is also equally important in addition to the El Nino-Southern Oscillation (ENSO), owing to the geographical location of Indian subcontinent. The hydroclimatic association of the variation of monsoon inflow into the Hirakud reservoir in India is investigated using ENSO and EQUatorial INdian Ocean Oscillation (EQUINOO, the atmospheric part of Indian Ocean Dipole mode) as the large-scale circulation information from tropical Pacific Ocean and Indian Ocean regions respectively. Individual associations of ENSO & EQUINOO indices with inflow into Hirakud reservoir are also assessed and found to be weak. However, the association of inflows into Hirakud reservoir with the composite index (CI) of ENSO and EQUINOO is quite strong. Thus, the large-scale circulation information from Indian Ocean is also important apart form the ENSO. The potential of the combined information of ENSO and EQUINOO for predicting the inflows during monsoon is also investigated with promising results. The results of this study will be helpful to water resources managers due to fact that the nature of monsoon inflow is becoming available as an early prediction.