Road-RFSense: A Practical RF-Sensing Based Road Traffic Estimation System for Developing Regions. Rijurekha Sen, Abhinav Maurya, Bhaskaran Raman, Amarjeet Singh, Rupesh Mehta, Ramakrishnan Kalyanaraman. ACM Transactions on Sensor Networks, 2014. [ abstract | pdf ]
An unprecedented rate of growth in the number of vehicles has resulted in acute road congestion problems worldwide, especially in many developing countries. In this article, we present Road-RFSense, a practical RF sensing--based road traffic estimation system for developing regions. Our first contribution is a new mechanism to sense road occupancy, based on variation in RF link characteristics, when line of sight between a transmitter-receiver pair is obstructed. We design algorithms to classify traffic states into two classes, free-flow versus congested, at timescales of 20 seconds with greater than 90% accuracy. We also present a traffic queue length measurement system, where a network of RF sensors can correlate the traffic state classification decisions of individual sensors and detect traffic queue length in real time. Deployment of our system on a Mumbai road gives correct estimates, validated against 9 hours of image-based ground truth. Our third contribution is a large-scale data-driven study, in collaboration with city traffic authorities, to answer questions regarding road-specific classification model training. Finally, we explore multilevel classification into seven different traffic states using a larger set of RF-based features and careful choice of classification algorithms.
KyunQueue: A Sensor Network System To Monitor Road Traffic Queues. Rijurekha Sen, Abhinav Maurya, Bhaskaran Raman, Rupesh Mehta, Ramakrishnan Kalyanaraman, Nagamanoj Vankadhara, Swaroop Roy, Prashima Sharma. 10th ACM Conference on Embedded Networked Sensor Systems, 2012 (acceptance rate: 18.7%). [ abstract | pdf ]
Unprecedented rate of growth in the number of vehicles has resulted in acute road congestion problems worldwide. Better traffic flow management, based on enhanced traffic monitoring, is being tried by city authorities. In many developing countries, the situation is worse because of greater skew in growth of traffic vs the road infrastructure. Further, the existing traffic monitoring techniques perform poorly in the chaotic non-lane based traffic here. In this paper, we present Kyun Queue, a sensor network system for real time traffic queue monitoring. Compared to existing systems, it has several advantages: it (a) works in chaotic traffic, (b) does not interrupt traffic flow during its installation and maintenance and (c) incurs low cost. Our contributions in this paper are four-fold. (1) We propose a new mechanism to sense road occupancy based on variation in RF link characteristics, when line of sight between a transmitter-receiver pair is obstructed. (2) We design algorithms to classify traffic states into congested or free-flowing at time scales of 20 seconds with above 90% accuracy. (3) We design and implement the embedded platforms needed to do the sensing, computation and communication to form a network of sensors. This network can correlate the traffic state classification decisions of individual sensors, to detect multiple levels of traffic congestion or traffic queue length on a given stretch of road, in real time. (4) Deployment of our system on a Mumbai road, after careful consideration of issues like localization and interference, gives correct estimates of traffic queue lengths, validated against 9 hours of image-based ground truth. Our system can provide input to several traffic management applications like traffic light control, incident detection, and congestion monitoring.
- Machine Learning Algorithms for Road Traffic State Classification in WirelessAcrossRoad. Abhinav Maurya. Master's Thesis, IIT Bombay.
Optimal Transport in Statistical Machine Learning: Selected Review and Some Open Questions. Abhinav Maurya. [ abstract | pdf ]
We review recent theoretical results underpinning the use of optimal transport and Wasserstein distances in statistical machine learning. The four primary results described in this report relate to the use of Wasserstein loss for train multi-class or multi-label classifiers, the use of composite Wasserstein loss to establish the convergence and contraction rates for mixing distributions in certain finite and infinite mixture models, the guarantees provided by optimal transport when used for domain adaptation, and the use of Wasserstein loss in improving the training procedure of Generative Adversarial Networks.
IEEE Big Data 2017 Panel Discussion on Bias and Transparency. Abhinav Maurya. AI Matters, Volume 4 Issue 2, July 2018. [ abstract | pdf ]
In January 2017, the ACM US Public Policy Council released a report on algorithmic transparency and accountability (ACM US Public Policy Council, 2017) which outlined several characteristics for algorithms to be considered transparent and accountable: (a) Awareness, (b) Access and redress, (c) Accountability, (d) Explanation, (e) Data Provenance, (f) Auditability, (g) Validation and Testing. A panel discussion on Big Data Bias and Transparency was organized at the IEEE International Conference on Big Data held in December 2017 to discuss opportunities and challenges faced by the data science community in their effort to incorporate the tenets of fairness, accountability, and transparency in their data-driven analyses and products. The panel consisted of Cynthia Dwork from Harvard University, John Langford from Microsoft Research, Jure Leskovec from Stanford University/Pinterest, Jeanna Matthews from Clarkson University, and was moderated by Ricardo Baeza-Yates from NTENT. This article provides an account of the panel discussion in the hope that it will be of interest to readers of AI Matters.
Data Center Systems: Recent Advances and Issues. Abhinav Maurya. [ abstract | pdf ]
Data center systems are at the heart of massive information systems like the Internet, ensuring availability and scaleability of services that demonstrate great variance in performance metrics. It is therefore necessary and interesting to study these systems in order to ensure that our data-intensive systems can fulfill the goals of efficiency, fault-tolerance, and scalability. We begin with the description of contemporary data centers as set forth in Kant's survey. We outline various advances and challenges in recent studies and practical deployments of data centers, in the areas of storage, networking, configuration management, and power management. We also describe the changing face of modern data centers from privately owned infrastructure to virtualized, geographically distributed and publicly leased commodity infrastructure. Next, we examine the various network topologies that have been proposed for data centers in recent literature. These include CamCube, DCell, and BCube. We conclude our studies in data center network topologies with a comparison delineated in Popa et al. We proceed to understand the transport layer issues in data centers. The TCP Incast problem is a particular problem that arises when using vanilla TCP inside data centers. We begin by understanding the causes and interactions involved in the occurrence of TCP Incast. We describe the solution to the TCP Incast problem proposed in Vasudevan et al. We conclude our study of transport layer issues in data centers by understanding DCTCP, a variant of TCP for data centers proposed in Alizadeh et al. In the final section of this study, we examine recent studies in automatic control of modern massive data centers that are not amenable to manual control. We outline the approach taken by Ganapathi et al. in automating the prediction of performance metrics for MapReduce jobs, and in their design of a realistic workload generator for testing MapReduce optimizations. Another notable work in this area is the online automation of resource allocation in data centers as described in Bodik et al. We conclude the report by listing some advances and issues that could not be covered in the report, and which form an active area of research in data center systems.