A5014551115- Advanced Wireless Network Optimization
- Quantum Computing Algorithms and Architecture
- Quantum Information and Cryptography
- Quantum Mechanics and Applications
- Advanced MIMO Systems Optimization
- Cooperative Communication and Network Coding
- Wireless Networks and Protocols
- Optimization and Search Problems
- Age of Information Optimization
- Energy Efficient Wireless Sensor Networks
- Advanced Bandit Algorithms Research
- Advanced Optimization Algorithms Research
- Complexity and Algorithms in Graphs
- IoT and Edge/Fog Computing
- Stochastic Gradient Optimization Techniques
- Interconnection Networks and Systems
- Mobile Ad Hoc Networks
- Network Traffic and Congestion Control
- Energy Harvesting in Wireless Networks
- Distributed Sensor Networks and Detection Algorithms
- Computational Drug Discovery Methods
- Electric and Hybrid Vehicle Technologies
- Advanced Queuing Theory Analysis
- Software-Defined Networks and 5G
- CCD and CMOS Imaging Sensors
IBM Research - Ireland
2023-2024
Yale University
2018-2022
Trinity College Dublin
2015-2019
National University of Ireland, Maynooth
2014
Universitat Pompeu Fabra
2012-2013
Federated learning (FL) allows model training from local data collected by edge/mobile devices while preserving privacy, which has wide applicability to image and vision applications. A challenge is that client in FL usually have much more limited computation communication resources compared servers a center. To overcome this challenge, we propose PruneFL -a novel approach with adaptive distributed parameter pruning, adapts the size during reduce both overhead minimize overall time,...
Federated learning (FL) allows model training from local data collected by edge/mobile devices while preserving privacy, which has wide applicability to image and vision applications. A challenge is that client in FL usually have much more limited computation communication resources compared servers a datacenter. To overcome this challenge, we propose PruneFL -- novel approach with adaptive distributed parameter pruning, adapts the size during reduce both overhead minimize overall time,...
We propose a channel access mechanism that allows Long Term Evolution (LTE) to operate in the 5-GHz unlicensed band (LTE-U/LAA) and fairly coexist with 802.11 Wireless Local Area Networks (WLANs). The proposed is compliant listen before talk, it can be configured maximize time used by LTE-U stations while coexisting WLANs. That is, an station will not affect throughput of WLAN more than if were station.
Abstract According to the ongoing IEEE 802.11ac amendment, wireless network is about embrace gigabit-per-second raw data rate. Compared with previous standards, this significant performance improvement can be attributed novel physical and medium access control (MAC) features, such as multi-user multiple-input multiple-output transmissions, frame aggregation, channel bonding. In paper, we first briefly survey main features of 802.11ac, then, evaluate these new in a fully connected mesh using...
Future mobile networks will exploit unlicensed spectrum to boost capacity and meet growing user demands cost-effectively. The 3 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">rd</sup> Generation Partnership Project (3GPP) has recently defined a License Assisted Access (LAA) scheme enable global Unlicensed LTE (U-LTE) deployment, aiming at 1) ensuring fair coexistence with incumbent WiFi networks, i.e., impacting on their performance no more...
Network densification over space and spectrum is expected to be key enabling the requirements of next generation mobile systems. The pitfall that radio resource allocation becomes substantially more complex. In this paper, we propose LaSR, a practical multi-connectivity scheduler for OFDMA-based multi-RAT LaSR makes optimal discrete control actions by solving sequence simple optimization problems do not require prior information traffic patterns. marked contrast previous work, flexibility...
We consider the proportional fair rate allocation in an 802.11 WLAN that supports multi-user MIMO (MU-MIMO) transmission by one or more stations. characterise, for first time, of MU-MIMO spatial streams and station opportunities. While a number features carry over from case without MU-MIMO, general neither flows nor stations need to be allocated equal airtime when is available.
In this paper, we bring the celebrated max-weight features (myopic and discrete actions) to mainstream convex optimization. Myopic actions are important in control because decisions need be made an online manner without knowledge of future events actions, many systems have a finite (so nonconvex) number decisions. For example, whether transmit packet or not communication networks. Our results show that these two can encompassed subgradient method for Lagrange dual problem by use stochastic...
This paper studies the problem of allocating band-width and computation resources to data analytics tasks in Internet Things (IoT) networks. IoT nodes are powered by batteries, can process (some of) locally, quality grade or performance how carried out depends on where these executed. The goal is design a resource allocation algorithm that jointly maximizes network lifetime subject energy constraints. joint maximization challenging with coupled constraints induce non-convexity. We first show...
We study the problem of operating a quantum switch with memory constraints. In particular, has to allocate memories clients generate link-level entanglements (LLEs), and then use these serve end-to-end requests. The paper's main contributions are (i) characterize switch's capacity region, (ii) propose allocation policy (MEW) that is throughput optimal. worst-case time complexity MEW exponential on system parameters. However, when requests bipartite LLE attempts always successful, we variant...
Recent advances in quantum computers are demonstrating the ability to solve problems at a scale beyond brute force classical simulation. As such, widespread interest algorithms has developed many areas, with optimization being one of most pronounced domains. Across computer science and physics, there number algorithmic approaches, often little linkage. This is further complicated by fragmented nature field mathematical optimization, where major classes problems, such as combinatorial convex...
We develop an edge-assisted object recognition system with the aim of studying system-level trade-offs between end-to-end latency and accuracy. focus on developing techniques that optimize transmission delay demonstrate effect image encoding rate neural network size these two performance metrics. explore optimal metrics by measuring our real time application. Our measurements reveal hitherto unknown parameter effects sharp trade-offs, hence paving road for optimizing this key service....
We study the problem of in-network execution data analytic services using multi-grade VNF chains. The nodes host VNFs offering different and possibly time-varying gains for each stage chain, our goal is to maximize analytics performance while minimizing transfer processing costs. VNFs' revealed only after their execution, since it data-dependent or controlled by third-parties, service requests network costs might also vary with time. devise an operation algorithm that learns, on fly, optimal...
We investigate the connections between max-weight approaches and dual subgradient methods for convex optimization. find that strong exist, we establish a clean, unifying theoretical framework includes both as special cases. Our analysis uses only elementary is not asymptotic in nature. It also allows us to an explicit direct connection discrete queue occupancies Lagrange multipliers.
Data centers are increasingly using high-speed circuit switches to cope with the growing demand and reduce operational costs. One of fundamental tasks is compute a sparse collection switching configurations support traffic matrix. Such problem has been addressed in literature variations approach proposed by Birkhoff 1946 decompose doubly stochastic matrix exactly. However, existing methods heuristic do not have theoretical guarantees on how well (i.e., permutations) can approximate scaled...
We show that the occupancy of appropriate queues can be used as a surrogate for Lagrange multipliers in convex optimisation. Our analysis uses only elementary methods, and is not asymptotic nature. One immediate consequence network problems scaled link queue when calculating dual function. Conversely, connection with casts light on behaviour under optimal decision-making (not just max-weight scheduling). Namely, links corresponding to active constraints necessarily grows step size α reduced....
Recent research on predicting the binding affinity between drug molecules and proteins use representations learned, through unsupervised learning techniques, from large databases of molecule SMILES protein sequences. While these have significantly enhanced predictions, they are usually based a limited set modalities, do not exploit available knowledge about existing relations among proteins. In this study, we demonstrate that by incorporating graphs diverse sources modalities into sequences...
This work describes the implementation of an Open Data platform fed by real-time data that are provided different city sensor networks. The aim paper is to show how enrich portals and require a new architecture deal with massive amounts continuously flowing information. introduces concept networks as key component in smart cities framework. Additionally, it deployment platform. Finally, discusses impact on both citizens services based potential mobile applications may be developed using this
We revisit the problem of traffic signal control in urban networks. Recent work literature has studied use scheduling algorithms from communication networks to with aim maximising network throughput. However, these are unable capture some characteristics such as protocol constraints or penalties on actions, e.g. loss time when changing between phases. In this paper, we propose a convex optimisation approach that decouples stability system choice policy. As result, able design range policies...
We propose an online policy that schedules the transmission and processing of data analytic tasks in Internet Things (IoT) network. The are executed with different precision at (possibly heterogeneous) nodes; network is subject to link bandwidth node capacity changes; task requests vary following unknown statistics. For this general IoT scenario, we formulate a resource allocation problem towards maximizing aggregate precision, design dynamic solution by combining FrankWolfe dual subgradient...
We consider the subgradient method for dual problem in convex optimisation with approximate multipliers, i.e., used update of variables is obtained using an approximation true Lagrange multipliers. This interesting problems where exact multipliers might not be readily accessible. For example, distributed available at nodes due to communication delays or losses. show that we can construct primal solutions get arbitrarily close set optima as step size α reduced. Applications analysis include...