Communication between neighboring drivers on the road is critical for safe and smooth driving. Drivers currently communicate by sounding horn or blinking lights, which sometimes is too coarse-grain to precisely convey the message, i.e., it is not apparent who the target and what the intent of the message is. Towards a better alternative, we explore the possibility of utilizing the smartphones of drivers to enable targeted communication between them. We propose a system, called DriverTalk, with which a driver can directly talk to convey messages such as you-are-blocking-traffic to the driver of the vehicle in front and do-not-tailgate-me to the driver behind. Furthermore, DriverTalk gives drivers the opportunity to seek and provide information as and when necessary during their driving. To identify the senders and receivers while exchanging messages, without an a priori knowledge about the drivers and their smartphones/vehicles, DriverTalk utilizes Visual IDs, i.e., the appearance images, of vehicles. Our evaluation shows that DriverTalk could deliver what a driver says to the target driver in about 1 second and with more than 95% success rate under various traffic conditions.

In recent years, there have been numerous geographically localized disruptions in Internet connectivity, as a result of malicious or natural causes. Though Internet-wide vulnerabilities have been studied extensively, the robustness of metropolitan area networks is still not well understood. In this paper, we study structural properties and vulnerabilities in metropolitan area networks, and develop techniques to monitor failures by small detection sets. By combining geolocation techniques with traceroute based measurements, we collect a large data set for 12 of the 25 most populated metropolitan regions in the US. Our analysis reveals that the structure of these paths gives superior insights into the robustness of metropolitan area networks as compared to the topological structure of the graph induced by these paths, which has been traditionally studied.

Performance optimization strategies in wireless mesh networks have witnessed many diverged directions, out of which the recent studies have explored the use of spatial multiplexing through data rate and transmit power adaptation to increase spatial re-usability while minimizing network interference. However, joint data rate and transmit power adaptation shows a clear trade-off, where higher transmit power helps in sustaining high achievable data rates with the cost of increased interference to the neighboring receiver nodes. Such a trade-off results in channel access unfairness among contending flows, where a node with high transmit power gains more performance benefit compared to its neighbors. Unfairness among nodes in a mesh network results in performance drops for the end-to-end flows. In this paper, we formulate a multivariate optimization problem to maximize network utilization while minimizing average transmit power of all transmitter nodes for successful communication. The Pareto optimality nature of the vector optimization has been explored to design a distributed localized heuristic where every node individually decides their scheduling slots and transmit power while keeping fairness as a constraint. The performance of the proposed scheme has been evaluated through simulation, and the comparisons with recent studies reveal that it results approximately 10%-40% improvement in end-to-end throughput for different network and traffic scenarios.

With advances in solar photovoltaic technology and decreasing costs of panels, solar energy is considered as an economical and sustainable solution for growing energy demands, especially in countries such as India with high insolation levels. However, key challenges exist in large scale adoption of distributed solar generation. Firstly, there are concerns related to breach of voltage regulations and system instabilities caused by distributed generation. Secondly, there are challenges in providing robust communication infrastructure for utilities to actively control and dispatch distributed generation. We address these challenges by designing a decentralised controller iPlug, which involves voltage-sensitive back-off of the level of solar capacity injection into the grid and is inspired by the CSMA protocol from Networking literature. This is done in conjunction with other resources like local batteries and local demand ramp-up resources. iPlug functions are complementary to emerging microinverter technologies and could in principle be incorporated into either the inverter controls or battery and home energy management control systems. We present preliminary results from simulation-based evaluation of iPlug algorithms and compare it with static PV injection strategies.

The tight coupling of computation, communication and control with physical systems such as actuation of closed-loop medical devices within the human body, peak power minimization by coordination of controllers across large industrial plants, and fast life-critical decision making by autonomous vehicles, present a set of fundamental and unique challenges. Each of these require new approaches at the intersection of multiple scientific, human and systems disciplines. We discuss five such challenges which require creative insights and application of model-based design, control systems, scheduling theory, formal methods, statistical machine learning and domain-specific experimentation. We ask the following questions: (1) An autonomous medical device is implanted to control your heart over a period of 5-7 years. How do you guarantee the software in the device provides safe and effective treatment under all physiological conditions? (2) Electricity prices in the US have summer peaks that are over 32x their average prices and winter peaks that are 86x. How can buildings respond to massive swings in energy prices at fast time scales? (3) While wireless has been successfully used for open-loop monitoring and tracking, how can we operate closed-loop control systems over a network of wireless controllers. Furthermore, how can we ensure robust, optimal and secure control in the presence of node/link failures and topology changes?

In order to reduce total energy communicated by distributed sensors, we address the problem where the recipient wants to perform supervised learning and classification on the data received from the sensors. Restricting our attention to a noiseless communication setting under simplistic Gaussian source assumptions, we provide algorithms for performing this supervised learning under energy limitations by modeling energy constraints through constraints on sum rate of communication from the sensors. We provide an algorithm for learning the distribution-parameters of sensor-data under rate constraints to arrive at high-reliability rate-allocation strategies, while enabling the rate-allocation algorithm to backtrack when the underlying distributions change, or when there is noise in sensed data that can push the algorithm towards a local minimum. Finally, we present numerical results on simulated data demonstrating the promise of the proposed techniques in reducing energy consumption.

The ubiquitous availability of image sensors in smartphones and other mobile computing devices makes feasible new applications for both barcodes that encode data as image patterns and watermarks that hide data in images. To support these new applications, a number of new methods have been developed for these long-standing image-based data communication interfaces, methods that are better suited for use with the image sensors. In this paper, we review these developments and the applications they enable, drawing upon examples from our own prior work in this area. We highlight key attributes that distinguish these image-based communications interfaces from other noncontact modes of communication and make their continued use an attractive option.

Direct measurements of the real world are typically expensive and time-consuming. Recent studies have revealed the promising quick alternative where real-world characteristics are indirectly estimated based on data readily available in the Internet. Such methods for indirect virtual-world estimation of real-world properties rely on the implicit hypothesis that the estimation accuracy depends on how extensively the Internet penetrates into the real world. Our paper makes the first step towards validating this hypothesis. We adopt a simple statistical model to quantify the relationship between an index of consumer prices and online searches for respective categories of goods and services across 19 countries. We also show how the strength of this relationship depends on Internet-penetration and other socioeconomic variables.

Many services, such as healthcare and education are highly human-intensive offerings that remain inaccessible (at acceptable quality level) to large numbers of people. With advances in computational power and increasing digitization of the world, there is an opportunity to apply data science to transform these services. We begin by describing a dire need and an opportunity to improve the healthcare system worldwide by supporting a shift from reactive treatment to more proactive action. As examples of what is possible, we present new machine learning techniques to predict a class of complications in an ICU, to identify patients in a hospital who are likely to require ICU admission, and to support triage of stroke patients. We also present preliminary work that shows the applicability of remote sensing and data analytics to measure body vitals such as respiration and heart rate, to screen for diseases, and to reduce the need for people to visit a hospital. We then describe a system called TutorSpace to help with personalization and improved navigation of videos from massive open online courses to enable more effective learning. Finally, we present data-driven techniques to improve public transportation services and to enable cities to reduce traffic congestion while offering a range of transportation options to their citizens. We frame all of the above efforts as examples of using data science to offer personalized services at scale. We describe some outstanding challenges that need to be met to achieve truly transformational impact.