A5061667884- Quantum Computing Algorithms and Architecture
- Complex Network Analysis Techniques
- Quantum Information and Cryptography
- Opinion Dynamics and Social Influence
- Age of Information Optimization
- Optimization and Search Problems
- Auction Theory and Applications
- Quantum Mechanics and Applications
- Advanced Wireless Network Optimization
- Advanced Memory and Neural Computing
- Advanced Bandit Algorithms Research
- Neural Networks and Applications
- Advanced MIMO Systems Optimization
- Wireless Communication Security Techniques
- Advanced Queuing Theory Analysis
- Low-power high-performance VLSI design
- Advancements in Semiconductor Devices and Circuit Design
- Distributed Sensor Networks and Detection Algorithms
- Evolutionary Game Theory and Cooperation
- Error Correcting Code Techniques
- Ferroelectric and Negative Capacitance Devices
- Advanced Wireless Communication Techniques
- Mobile Crowdsensing and Crowdsourcing
- Social Media and Politics
- Opportunistic and Delay-Tolerant Networks
Indian Institute of Technology Madras
2016-2025
Intel (United States)
2023
University of Illinois Urbana-Champaign
2014-2021
The University of Texas at Austin
2012-2015
General Electric (United States)
2002-2003
This paper presents a high-yielding backside power delivery (BPD) technology, PowerVia, implemented on Intel 4 finFET process. PowerVia more directly integrates to the transistor as compared published buried rail schemes, enabling additional wiring resources front side for signal routing. A fabricated E-core with $\gt 90$% cell utilization showed 30$% platform voltage droop improvement and 6% frequency benefit similar design without PowerVia. Transistor performance, reliability, fault...
Unlike traditional graph-based linear dynamics, where agents exchange opinions with their neighbors in a static social graph regardless of differences opinions, the bounded confidence opinion dynamics models between similar opinions. We generalize model by incorporating pairwise stochastic interactions, probabilistic influencing based on and self or endogenous evolutions agent which are represented random processes. The exchanges resulting from have contraction effects, whereas motions an...
We study the information-theoretic limit of reliable information processing by a server with queue-length dependent quality service. define capacity for such system as number bits reliably processed per unit time, and characterize it in terms queuing parameters. also distributions arrival service processes that maximize minimize systems discrete-time setting. For at most one time slot, we observed minimum around memoryless distribution. studied case multiple arrivals burstiness has adverse...
We study the effect of external infection sources on phase transitions in epidemic processes. In particular, we consider an spreading a network via SIS/SIR dynamics, which addition is aided by agents - unconstrained graph, but possessing limited rate or virulence. Such model captures many existing models externally epidemics, and finds use settings epidemiology, marketing advertising, robustness, etc. provide detailed characterization impact thresholds. for SIS model, show that any strategy...
Many companies now use crowdsourcing to leverage external (as well as internal) crowds perform specialized work, and so methods of improving efficiency are critical. Tasks in systems with work have multiple steps each step requires skills. Steps may different flexibilities terms obtaining service from one or agents, due varying levels dependency among parts steps. a task precedence constraints them. Moreover, there variations loads types tasks requiring skill-sets availabilities agents...
Traditional models in opinion dynamics involve agents updating their opinions based on the of neighbors a static social-graph, regardless differences opinions. In contrast, bounded confidence does not presume interaction graph, and instead interactions between those that share similar (i.e., are close to one another, capturing online discussion groups conventional meetings). We generalize model by incorporating pairwise stochastic as well self or endogenous evolution agent opinions, which is...
We consider a setting where qubits are processed sequentially, and derive fundamental limits on the rate at which classical information can be transmitted using quantum states that decohere in time. Specifically, we model sequential processing of single server queue, explicit expressions for capacity such `queue-channel.' also demonstrate sweet-spot phenomenon with respect to arrival i.e., show there exists value transmission (in bits/sec) through queue-channel is maximised. Next, average...
Due to the unprecedented success of deep neural networks in inference tasks like speech and image recognition, there has been increasing interest using them mobile in-sensor applications. As most current are very large size, a major challenge lies storing network devices with limited memory. Consequently is growing compressing by quantizing synaptic weights, but prior work heuristic lacking theoretical foundations. Here we develop an approach functional high-rate quantization theory. Under...
Crowdsourcing of jobs to online freelance markets is rapidly gaining popularity. Most crowdsourcing platforms are uncontrolled and offer freedom customers freelancers choose each other. This works well for unskilled (e.g., image classification) with no specific quality requirement since functionally identical. For skilled software development) requirements, however, this does not ensure the maximum number job requests satisfied. In work we determine capacity markets, in terms satisfied...
In a vast body of opinion dynamics literature, an agent updates its based on the opinions neighbors in static social graph, regardless their differences opinions. contrast, bounded confidence does not presume interaction and instead limits interactions to those agents that share related (i.e., whose are close one another). We generalize model by incorporating stochastic endogenous evolution opinions, which itself is random process. analytically characterize conditions under this stable...
Recent trends suggest that cognitive radio based wireless networks will be frequency agile and the nodes equipped with multiple radios capable of tuning across large swaths spectrum. The MAC scheduling problem in such refers to making intelligent decisions on which communication links activate at time instant over band. challenge designing a low-complexity distributed MAC, achieves low delay, is posed by two additional dimensions networks: interference graphs data rates are frequency-band...
As techniques for fault-tolerant quantum computation keep improving, it is natural to ask: what the fundamental lower bound on space overhead? In this paper, we obtain a overhead required <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mi>&#x03F5;</mml:mi></mml:math>-accurate implementation of large class operations that includes unitary operators. For practically relevant case sub-exponential depth and sub-linear gate size, our tighter than known bounds. We by connecting...
New device technologies such as spintronics, carbon nanotubes, and nanoscale CMOS incur random transient failures, where the failure probability is governed by energy consumption through energy-failure functions. At same time, there growing use of deep neural networks for many inference applications, specialized hardware being developed with these nanotechnologies physical substrates. It important to understand basic energy-reliability limits. Using Pippenger's mutual information propagation...
Many companies now use crowdsourcing to leverage external as well internal crowds perform specialized work, and so methods of improving efficiency are critical. Tasks in systems with work have multiple steps each step requires skills. Steps may different flexibilities terms obtaining service from one or agents due varying levels dependency among parts steps. a task precedence constraints them. Moreover, there variations loads types tasks requiring skill sets availabilities sets. Considering...
In the nanoscale regime, behavior of both extant and emerging semiconductor devices are often unreliable. Reliability such trades-off with their energy consumption, speed, and/or chip area. We study reliability-energy limits for circuits designed using devices; examples drawn from spin electronics. Using mutual information propagation in logic technique developed by Pippenger, together optimization, we obtain lower bounds on consumption computing n-input Boolean functions. observe that...
Crowdsourcing of jobs to online freelance platforms is rapidly gaining popularity. Most crowdsourcing are uncontrolled and offer freedom customers freelancers choose each other. This works well for unskilled (e.g., image classification) with no specific quality requirement since functionally identical. For skilled software development) requirements, however, this does not ensure that the maximum number job requests satisfied. In paper, we determine capacity regulated systems, in terms...
The synthesis and testing issues for certain nonclassical faults (delay bridging faults) are considered. Cost versus performance tradeoffs in synthesizing circuits which robust delay fault testable first examined. A new type of scan latch design is presented testability sequential circuits. combinational logic also addressed. Tests generated using implicit don't cares obtained the circuit during process. PODEM-based algorithm has been developed fast detection a class faults.< <ETX...
Due to process variation in nanoscale manufacturing, there may be permanently missing connections information processing hardware. timing errors circuits, missed messages intra-chip communications, equivalent transiently connections. In this work, we investigate the performance of message-passing LDPC decoders presence We prove concentration and convergence theorems that validate use density evolution analysis. Arbitrarily small error probability is not possible with connections, but find...
We consider a queue-channel model that captures the waiting time-dependent degradation of information bits as they wait to be transmitted. Such scenario arises naturally in quantum communications, where tend decohere rapidly. Trailing capacity results obtained recently for certain queue-channels, this paper aims construct practical channel codes erasure (EQC)—a characterized by highly correlated erasures, governed underlying queuing dynamics. Our main contributions are twofold: (i) propose...
We consider a setting where stream of qubits is processed sequentially. derive fundamental limits on the rate at which classical information can be transmitted using that decohere as they wait to processed. Specifically, we model sequential processing single server queue, and expressions for capacity such quantum `queue-channel.' Focusing erasures, obtain an explicit single-letter formula in terms stationary waiting time queue. Our proof also implies `classical' coding/decoding strategy...
We propose a methodology to perform process variation-aware device and circuit design using fully physics-based simulations within limited computational resources, without developing compact model. Machine learning (ML), specifically support vector regression (SVR) model, has been used. The SVR model trained dataset of devices simulated priori, the accuracy prediction by demonstrated. To produce switching time distribution from ML we only had generate train validate which needed ∼500 hours...
This paper considers a natural and widely prevalent setting where collection of one dimensional time series evolve in causal manner, is interested inferring the graph governing causality between these processes high setting. We consider this problem special case variables are discrete updates Markov. develop new algorithm to learn structure based on notion directed information, analytically empirically demonstrate its performance. Our an adaptation greedy heuristic for learning undirected...