Jekyll2019-07-17T11:08:36+09:00http://localhost:4000/abacus/feed.xmlABACUS ProjectWebsite for the project Adaptive ambient BAckscatter Communications for Ultra-low power Systems (ABACUS)Learning with Vertically-Partitioned Data, Binary Feedback, and Random Parameter Update2019-06-26T00:00:00+09:002019-06-26T00:00:00+09:00http://localhost:4000/abacus/2019/06/26/vertially-partioned<p>Ngu Nguyen presented our paper at Workshop on Hot Topics in Social and Mobile Connected Smart Objects.
The workshop focuses on experiences with the design, implementation, deployment, operation, and evaluation of novel systems for smart objects and the social aspects of these systems in the emerging cooperative environments.</p>
<h4 id="abstract">Abstract</h4>
<p>Machine learning models can deal with data samples scattered among distributed agents, each of which holds a non-overlapping set of sample features.
In this paper, we propose a training algorithm that does not require communication between these agents.
A coordinator can access ground-truth labels and produces binary feedback to guide the optimization process towards optimal model parameters.
We mimic the gradient descent technique with information observed locally at each agent.
We experimented with the logistic regression model on multiple benchmark datasets and achieves promising results in terms of convergence rate and communication load.</p>Ngu Nguyen presented our paper at Workshop on Hot Topics in Social and Mobile Connected Smart Objects. The workshop focuses on experiences with the design, implementation, deployment, operation, and evaluation of novel systems for smart objects and the social aspects of these systems in the emerging cooperative environments.Adaptive ambient BAckscatter Communications for Ultra-low power Systems (ABACUS)2019-05-20T00:00:00+09:002019-05-20T00:00:00+09:00http://localhost:4000/abacus/2019/05/20/project-information<h4 id="adaptive-ambient-backscatter-communications-for-ultra-low-power-systems-abacus">Adaptive ambient BAckscatter Communications for Ultra-low power Systems (ABACUS)</h4>
<p>Low power wireless communications has been identified as one of the key enabling technologies for the Internet of Things (IoT). The performance of the contemporary IoT connectivity solutions are mainly limited by the two main factors: congestion and interference. In particular, these degrading factors are more stringent in the license-exempt frequency bands. Together with the limited operation time of the battery-powered devices, they hamper the scaling of the IoT deployments. This proposal seeks to overcome these limitations and allow the IoT devices to achieve over 15-year battery operation time by utilizing ambient backscatter communications (AmBC).
AmBC is a recently proposed ultra-low power communication scheme in which the communicating device modulates the ambient radio signal impinging at its antennas. It does not need a power hungry transceiver and achieves several orders of magnitude lower power consumption than current radio modems. This reduced power consumption directly translates to longer battery operation time.
Alternatively, the AmBC can be combined with an energy harvester to allow energy autonomous operation. The AmBC systems proposed in the literature so far have relayed on relatively simple single antenna designs that cannot be adapted to the spatio-temporal variations of the ambient source signals.
In other areas of wireless communications, the use of multiple antennas has become mainstream. In this project we seek to utilize smart, adaptable antennas and co-design and optimize the device and modulator, utilized waveform and receiver algorithms as well as the sensor applications in order to significantly improve the system performance in terms of energy efficiency, spectrum efficiency and communications range.</p>Adaptive ambient BAckscatter Communications for Ultra-low power Systems (ABACUS)