Ankan Mullick, Sombit Bose, Rounak Saha, Ayan Kumar Bhowmick, Aditya Vempaty, Prasenjit Dey, Ravi Kokku, Pawan Goyal, Niloy Ganguly. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

With the recent advances in Internet of Things, smarthomes containing a variety of sensors are expected to monitor and react to the activities of daily living (ADL) of the residents with the goal of improving convenience, comfort, and safety. However, a single human activity may trigger multiple sensors, and each sensor is triggered by multiple activities. Also, the very human nature of interleaving activities along with the presence of multiple inhabitants makes ADL detection a complex problem. This paper tackles the problem with a data driven de-multiplexing approach termed TSFS (Temporal-Sensor-Frequency-Stitch) that disentangles each individual activity from the sensor stream and thus simplifies the ADL recognition problem. TSFS leverages temporal(T), sensory(S), and frequency(F) information to discriminate between two consecutive/parallel activities, followed by a stitch(S) to enhance segmentation. The multi-faceted approach helps us in correctly identifying segments; extensive evaluation on 10 public datasets reveals that TSFS is better by 185 percent when compared to a simple temporal segmentation approach. The activity detection accuracy yields within 93 percent of the best possible (an oracle) approach.

Intelligent tutoring systems (ITS) have been a topic of great interest for about five decades. Over the years, ITS research has leveraged AI advancements, and has also helped push the boundaries of AI capabilities with grounded usage scenarios. Using ITSs along with classroom instruction to augment traditional teaching is a canonical example of how humans and machines can work together to solve problems that are otherwise overwhelming and non-scalable individually. The experiences of personalized learning created by (1) seamless orchestration of human decision-making at few critical points with (2) scalability of cognitive capabilities using AI systems can drive increased student engagement leading to improved learning outcomes. By considering two particular use-cases of early childhood learning and higher education, we discuss the challenges involved in designing these complex human-centric systems. These systems integrate technologies involving interactivity, dialog, automated question generation, and learning analytics.

Powered by the emergence of the Internet of Things, smart homes containing a variety of sensors and actuators are expected to monitor and react to the activities of the residents with the goal of improving convenience, comfort and safety. However, in typical home settings, each human Activity of Daily Living (ADL) generates events from multiple sensors, and each sensor is triggered by multiple ADLs. Consequently, achieving high detection accuracy in these complex environments requires large amounts of training data for every possible multiplexing scenario, making it a complex problem. In this paper, we propose a data driven three-step de-multiplexing approach that simplifies the ADL recognition problem by first segmenting the event stream into periods of interest, before feeding to a classifier. We mine datasets to identify salient features which allow us to achieve a good segmentation. Extensive evaluation on ten public datasets shows that our approach achieves upto 77% segmentation accuracy, and a activity detection accuracy within 91% of the best possible.

IoT enabled smart environments are expected to proliferate significantly in the near future, particularly in the context of monitoring services for wellness living, patient healthcare and elderly care. Timely maintenance of failed sensors is of critical importance in such deployments to ensure minimal disruption to monitoring services. However, maintenance of large and geographically spread deployments can be a significant challenge. We present Idea that significantly increases the vtime-before-repair for a smart home deployment, thereby reducing the maintenance overhead. Specifically, our approach leverages the facts that (a) there is inherent sensor redundancy when combinations of sensors monitor activities of daily living (ADLs) in smart environments, and (b) the impact of each sensor failure depends on the activities being monitored and the functional redundancy afforded by rest of the heterogeneous sensors available for detecting the activities. Consequently, Idea identifies homes that need to be fixed based on expected degradation in ADL detection performance, and optimizes maintenance scheduling accordingly. We demonstrate that our approach leads to 3--40 times fewer maintenance personnel than a scheme in which failed sensors are fixed without considering their impact.

Recent studies have shown that mobile applications are increasingly becoming "chatty", resulting in excessive signaling load on cellular networks, potentially causing major network outages. One of the main causes of this signaling overload is the static dormancy timer that 3GPP-based networks employ to control the release of radio resources allocated to a device. To systematically manage the ever-increasing signaling load, we design a novel network-aware system named REACT for dynamically tuning the dormancy timers. In doing so, we highlight a key trade-off between signaling overhead on the network and energy consumed on mobile devices. Contrary to the prior work, our solution models and predicts traffic of mobile apps at the network side with high degree of accuracy, and then dynamically tunes the dormancy timer at a per-device, per-base-station or per-RNC granularity, the approach requires no modifications on mobile devices. Our evaluation shows that REACT reduces signaling overhead by more than a factor of 2 compared to current static setting of dormancy timers. Furthermore, network-aware dynamic tuning of dormancy timers enables a flexible system that can aggressively save energy on devices when the overall network load is low, and manages signaling overhead effectively during high network load.

We handle the problem of efficient user-mobility driven macro-cell planning in cellular networks. As cellular networks embrace heterogeneous technologies (including long range 3G/4G and short range WiFi, Femto-cells, etc.), most traffic generated by static users gets absorbed by the short-range technologies, thereby increasingly leaving mobile user traffic to macro-cells. To this end, we consider a novel approach that factors in the trajectories of mobile users as well as the impact of city geographies and their associated road networks for macro-cell planning. Given a budget k of base-stations that can be upgraded, our approach selects a deployment that improves the most number of user trajectories. The generic formulation incorporates the notion of quality of service of a user trajectory as a parameter to allow different application-specific requirements, and operator choices. We show that the proposed trajectory utility maximization problem is NP-hard, and design multiple heuristics. We evaluate our algorithms with real and synthetic datasets emulating different city geographies to demonstrate their efficacy. For instance, with an upgrade budget k of 20%, our algorithms perform 3-8 times better in improving the user quality of service on trajectories when compared to greedy location-based base-station upgrades.

Over-the-top estimation of available bandwidth (AB) in a network path has been well studied for wired networks. The AB of a path denotes its slack capacity, i.e., the bandwidth available for use in the path without impacting the existing traffic. This estimation problem has been receiving attention only recently in cellular networks, which are increasingly becoming one of the main modes of access for a large number of applications. In this paper, we discuss the challenges posed by the problem, and why existing techniques developed for wired networks cannot be applied. We show that, interestingly, it may not even be feasible to estimate AB using over-the-top approaches under certain conditions, even when the wireless channel and traffic conditions are non-varying. We then present a novel AB estimation technique for cellular networks, which typically use proportional fair scheduling at base stations. When the wireless channel and traffic conditions are non-varying, our technique can accurately determine AB when it exceeds the “fair share” due to a new flow. We also extend the basic technique for estimation under conditions that are time-varying. The proposed methods can as well be used when one or more bottleneck links in a network path are fair-scheduled using algorithms such as weighted-fair queueing. We evaluate our methods using simulations and over operational networks, and present the results. In simulations, our technique is capable of detecting AB close to 90% of the time under feasible conditions even with bursty traffic.

We conduct an extensive measurement study in an operational cellular network in dense urban settings and identify two key sources of inefficiencies in standard 3GPP handoff technique: wrong timing of handoff and sub-optimal base station association at a point. Based on our study, we propose a novel handoff technique that leverages (1) historical geo-tagged signal strength data from visible base stations at typical mobile user locations and (2) knowledge of the user trajectory. First, we formulate the handoff problem as a bi-criteria optimization problem and present a dynamic programming formulation that provides pareto-optimal solutions that maximize overall signal strength along user trajectory for a range of handoffs. Using an efficient utility based selection technique, we select a solution that balances the tradeoff between reducing number of handoffs and increasing the overall signal strength experienced by the user. Next, we present a much faster greedy heuristic that augments the solution obtained from the dynamic programming formulation when optimizing user performance in the presence of dynamic network load at each base station. A key contribution of our work is to show that striking a tradeoff between the two objectives is much harder in dense macro cellular base station deployments and an effective handoff technique should balance both these objectives well. We show that our technique takes a well balanced approach and achieves more than 25% reduction in the number of handoffs over the standard 3GPP handoff technique in an operational cellular network without incurring reduction in average signal strength, and even improving it by 1dB/sec in some cases. Finally, we propose a novel 3GPP standards compliant trajectory-aware location based handoff protocol and discuss the practical implementation aspects in cellular networks.

We design and implement a novel system called Async, which enables a mobile network operator (MNO) to efficiently manage the growth of mobile data by leveraging the delay-elastic nature of certain applications and the price-sensitive nature of certain users. Specifically, Async introduces an alternate “asynchronous” content-delivery paradigm for heavy content (e.g., videos), and facilitates an MNO to negotiate with users a delay in delivery in exchange for appropriate incentives. The MNO uses the negotiated delays to actively manage Async flows to reduce congestion and improve the quality-of-experience (QoE) of both delayed and regular flows. We show that in comparison to state-of-the-art, Async's network-based flow management enhances QoE for more than 30% of the regular flows, with up to 60% improvement in per-flow QoE metric, while still meeting the negotiated delivery times of 95% of the delayed flows. Async also lowers the delivery times of delayed flows by ∼67% and significantly increases robustness to traffic unpredictability. Our design is robust to disconnections and does not require any modifications to existing network infrastructure and protocols. Our prototype deployment (using Apache's mod_proxy and an Android app) on live networks confirms Async's efficacy in meeting EDTs for diverse deployment scenarios.

In this work, we demonstrate the design and implementation of a novel system called Async, which enables a mobile network operator (MNO) to efficiently manage the growth of mobile data. For this, we enable an opportunistic delivery model for the MNOs by leveraging the delay-elastic nature of certain applications such as video downloads, and the price-sensitive nature of certain users. Specifically, in Async, we enable an alternate content delivery paradigm that facilitates an MNO to control when, and at what price, the content is delivered to the users. In our system, users can select a price and expected delivery time (or EDT) option from a multitude of choices. We develop a prototype of such a system by extending Apache's mod_proxy module and by developing a service on the Android platform without requiring any modifications to existing network infrastructure or protocols. We demonstrate two applications using such a service: (1) download of large files, e.g., software updates, movie files, and (2) an educational App that downloads educational content on mobile devices.

The Internet is witnessing a rapid increase in video traffic. Due to the scalability and the cost-savings offered by cloud-computing, Internet video service providers are increasingly delivering their content from multi-tenant cloud data centers. One of the major challenges faced by such a video service provider is the management of the Quality-of-Experience (QoE) of the end-users in the presence of Variable Bit Rate (VBR) video flows, time varying network conditions in the Internet, and the bounded egress bandwidth provided by the data center. To this end, we present InSite, a light-weight and easy-to-deploy solution for managing the QoE of a set of video flows of a service provider, which are served from a data center. InSite is deployed at the egress of a data center, between the video servers and the clients, and manages the video flows that are transmitted over TCP. The solution uses a novel generalized binary search technique to concurrently search for the appropriate flow rates for a set of flows, with the goal of maximizing the QoE-fairness across the flows, as opposed to TCP-fairness. The search takes into account the total egress bandwidth allocated for the set of video flows at the data center, the unknown and possibly time-varying capacities of any remote bottleneck links, and the playout buffer sizes of the video flows. The solution is also designed to operate with minimal modifications to the video servers and the clients. In our evaluations using extensive ns-3 simulations and a testbed implementation for serving videos over TCP, we observe that deploying InSite achieves several folds reduction in playout stalls over a system without InSite.

This paper describes the design and implementation of a network virtualization substrate (NVS) for effective virtualization of wireless resources in cellular networks. Virtualization fosters the realization of several interesting deployment scenarios such as customized virtual networks, virtual services, and wide-area corporate networks, with diverse performance objectives. In virtualizing a base station's uplink and downlink resources into slices, \ssrNVSmeets three key requirements-isolation, customization, and efficient resource utilization-using two novel features: 1)NVSintroduces a provably optimal slice scheduler that allows existence of slices with bandwidth-based and resource-based reservations simultaneously; and 2)NVSincludes a generic framework for efficiently enabling customized flow scheduling within the base station on a per-slice basis. Through a prototype implementation and detailed evaluation on a WiMAX testbed, we demonstrate the efficacy of \ssrNVS. For instance, we show for both downlink and uplink directions that \ssrNVScan run different flow schedulers in different slices, run different slices simultaneously with different types of reservations, and perform slice-specific application optimizations for providing customized services.

This paper is motivated by two observations: (1) cellular network operators are actively exploring advertisement delivery as a new means of revenue generation, and (2) cellular basestations perceive intermittent overloads at different times of day. Bringing the two observations together, we design and implement Opal, a novel system for opportunistically aligning advertisement delivery with basestation overload. Such alignment improves the overall perception of network availability for users. To achieve the alignment systematically, Opal builds on an analytical framework for tunable unavailability of network service to users during overload. At the same time, if the network is not overloaded enough during a certain period, Opal schedules enough advertisements to satisfy the advertisement delivery contracts. Opal minimizes the amount of state to be maintained to play advertisements to users uniformly and also maximizes the number of viewers for each advertisement. We implement a prototype of Opal on a Picochip based WiMAX testbed, and demonstrate its efficacy using simulations, analysis and prototype evaluation.

We demonstrate OPAL [2]- a system for opportunistically aligning advertisement delivery with basestation overload periods, which reduces the total number of times network service is unavailable to users. This system is motivated by two trends: Firstly, as networks get dominated by non-elastic traffic such as video, overload management using admission control will be imminent, thereby exposing network unavailability to users. Secondly, as mobile advertising is rapidly growing, mobile network operators are increasingly interested in leveraging its huge revenue potential, and are also better equipped than any other entity in reaching the users.

This paper describes the design and implementation of a network virtualization substrate NVS) for effective virtualization of wireless resources in WiMAX networks. Virtualization fosters the realization of several interesting deployment scenarios such as customized virtual networks, virtual services and wide-area corporate networks, with diverse performance objectives. In virtualizing a basestation's uplink and downlink resources into slices, NVS meets three key requirements - isolation, customization, and efficient resource utilization - using two novel features: (1) NVS introduces a provably-optimal slice scheduler that allows existence of slices with bandwidth-based and resource-based reservations simultaneously, and (2) NVS includes a generic framework for efficiently enabling customized flow scheduling within the basestation on a per-slice basis. Through a prototype implementation and detailed evaluation on a WiMAX testbed, we demonstrate the efficacy of NVS. For instance, we show for both downlink and uplink directions that NVS can run different flow schedulers in different slices, run different slices simultaneously with different types of reservations, and perform slice-specific application optimizations for providing customized services.

Adaptive transmit power control in 802.11 wireless LANs (WLANs) on a per-link basis helps increase network capacity and improves battery life of WiFi-enabled mobile devices. However, it faces the following challenges: 1) it can exacerbate receiver-side interference and asymmetric channel access; 2) it can incorrectly lead to lowering the data rate of a link; 3) mobility-induced channel variations at short timescales make detecting and avoiding these problems more complex. Despite substantial prior research, state-of-the-art solutions lack comprehensive techniques to address the above problems. In this paper, we design and implement Symphony, a synchronous two-phase rate and power control system whose agility in adaptation enables us to systematically address the three problems while maximizing the benefits of power control on a per-link basis. We implement Symphony in the Linux MadWifi driver and show that it can be realized on hardware that supports transmit power control with no modifications to the 802.11 MAC, thereby fostering immediate deployability. Our extensive experimental evaluation on a real testbed in an office environment demonstrates that Symphony : 1) enables up to 80% of the clients in three different cells to settle at 50%-94% lower transmit power than a per-cell power control solution; 2) increases network throughput by up to 50% across four realistic deployment scenarios; 3) improves the throughput of asymmetry-affected links by 300%; and 4) opportunistically reduces the transmit power of mobile clients running VOIP calls by up to 97% while only causing a negligible degradation of voice quality.

Several research works have argued that adaptive beamforming has the potential to realize the high spectral efficiency requirements of next-generation wireless standards, and is especially well-suited for multipath-rich environments such as indoors. Most works have been limited to theory; few works in literature address the practical benefits and realizability of adaptive beamforming. In this paper, we design and implement the first indoor WLAN beamforming system with multi-element array antennas and software radio platforms, that forms a testbed for exploration of practical benefits of beamforming, and evaluation of algorithms for efficient beamforming in diverse environments. In the process of building the system, we identify and address several challenges with practical beamforming that are often ignored in theoretical works. Most importantly, channel estimation for forming the best beam to a user is hindered by oscillator drifts on the transmitter and receiver side that introduce hard-to-isolate phase and frequency offsets from the estimated channel coefficients. We describe these issues and incorporate novel solutions in our system to address them without requiring hardware modifications. We use the system to demonstrate the realizable benefits of adaptive beamforming in a typical indoor office environment.

We design, implement, and evaluate a vehicular communication system that improves uplink connectivity through multi-lobe beam pattern switching on a smart antenna. Directionality and base station-diversity are two well-known, independently developed mechanisms for improving the uplink connectivity of mobile clients. In this paper, we highlight that a system combining both mechanisms can achieve significant improvement in performance with multi-lobe beams that strike a tradeoff between directionality and diversity. This is in contrast to the mere steering of narrow beams used in conventional smart antenna systems. For tractability at vehicular speeds, our R2D2 system searches through a limited set of beam patterns with different numbers of lobes, and includes a two-stage algorithm that uses both runtime adaptation and cached candidate patterns. We design and evaluate several variants of run-time adaptation that tune the number and angle of lobes in the beam, and the bit rate. The design of these algorithms is guided by both analysis and real-world measurements with a smart antenna system mounted on a vehicle. These measurements with our prototype implementation show that R2D2 can achieve an uplink throughput increase of up to 154% over pure beamsteering and 45% over pure basestation diversity.

In this paper, we address the placement of relay service agents (routelets) in the Internet to assist multipath transport protocols; these protocols are designed to achieve better network utilization and fairness by exploiting path diversity in the Internet. We identify three different routelet deployment scenarios, provide LP formulations for routelet placement in each of the scenarios, and prove that the placement problem is NP-hard in two of the scenarios. We provide rounding algorithms with provable properties for a subset of the scenarios, and compare their performance through simulations on several BRITE topologies of varying scales. We observe that our rounding algorithm leads to a 2-5 fold increase in bandwidth utilization compared to the default single path routing. This closely matches the performance of the optimal LP throughput in the several topologies tested.

Sophisticated middlebox services-such as network monitoring and intrusion detection, DDoS mitigation, worm scanning, XML parsing and protocol transformation-are becoming increasingly popular in today's Internet. To support high- throughput, these services are often deployed on distributed memory, multi-processor (DM-MP) hardware platforms such as a cluster of network processors. Scaling the throughput of such platforms, however, is challenging because of the difficulties and overheads of accessing persistent, shared state maintained by the services. In this paper, we describe the design and implementation of Oboe, a run-time system for DM-MP platforms that addresses the above challenge through two foundations: (1) category-specific management of shared state, and (2) adaptive flow- level load distribution for addressing persistent processor overload. Our simulations demonstrate that Oboe can achieve performance within 0-5% of an ideal adaptive system. Our prototype implementation of Oboe on a cluster of IXP2400 network processors, demonstrates the scalability achieved with increasing number of processors, number of flows and state size.

Adaptive transmit power control in 802.11 Wireless LANs (WLANs) on a per-link basis helps increase network capacity and improves battery life of Wifi-enabled mobile devices. However, it faces the following challenges: (1) it can exacerbate receiver-side interference and asymmetric channel access, (2) it can incorrectly lead to lowering the data rate of a link, (3) mobility-induced channel variations at short timescales make detecting and avoiding these problems more complex. Despite significant research in rate and power control, state of the art solutions lack comprehensive techniques to address the above problems.In this paper, we design and implement Symphony - a Synchronous Two-phase Rate and Power control system, whose agility in adaptation enables us to systematically address the three problems, while maximizing the benefits of power control on a per-link basis. We implement Symphony in the Linux Madwifi driver, and show that it can be realized on hardware that supports transmit power control with no modifications to the 802.11 MAC, thereby fostering immediate deployability. Our extensive experimental evaluation on a real testbed in an office environment demonstrates that Symphony (1) enables up to 80% of the clients in 3 different cells to settle at 50% to 94% lower transmit power than a per-cell power control solution, (2) increases network throughput by up to 50% across realistic deployment scenarios, (3) improves the throughput of asymmetry-affected links by 300%, and (4) opportunistically reduces the transmit power of mobile clients running VOIP calls by up to 97%, while causing minimum impact on voice quality.

Background transfers, or transfers that humans do not actively wait on, dominate the Internet today. In today's best-effort Internet, background transfers can interfere with foreground transfers causing long wait times, thereby hurting human productivity. In this paper, we present the design and implementation of a background network, Harp, that addresses this problem. Harp has three significant advantages over recent end-host based background transport protocols; Harp (i) uses multiple paths to exploit path diversity and load imbalance in the Internet to tailor network resource allocation to human needs, (ii) provides better fairness and utilization compared to unipath end-host protocols, and (ill) can be deployed at either end-hosts or enterprise gateways, thereby aligning the incentive for deployment with the goals of network customers. Our evaluation using simulations and a prototype on Planetlab suggests that Harp improves foreground TCP transfer time by a factor of five and background transfer time by a factor of two using just two extra paths per connection.

In this paper, we describe the design and evaluation of a scheduler (referred to as Everest) for allocating processors to services in high performance, multi-service routers. A scheduler for such routers is required to maximize the number of packets processed within a given delay tolerance, while isolating the performance of services from each other. The design of such a scheduler is novel and challenging because of three domain-specific characteristics: (1) difficult-to-predict and high packet arrival rates, (2) small delay tolerances of packets, and (3) significant overheads for switching allocation of processors from one service to another. These characteristics require that the scheduler be agile and wary simultaneously. Whereas agility enables the scheduler to react quickly to fluctuations in packet arrival rates, wariness prevents the scheduler from wasting computational resources in unnecessary context switches. We demonstrate that by balancing agility and wariness, Everest, as compared to conventional schedulers, reduces by more than an order of magnitude the average delay and the percentage of packets that experience delays greater than their tolerance. We describe a prototype implementation of Everest on Intel's IXP2400 network processor

We present half-pipe anchoring, a technique that enables multiple connection handoff mechanisms that are efficient and easy to implement. In a server cluster, these mechanisms result in better resource utilization and improved scalability. More importantly, half-pipe anchoring supports the only connection handoff mechanism that operates efficiently in heterogeneous clusters composed of specialized nodes. The key idea behind our approach is to decouple the two unidirectional half-pipes that make up a TCP connection between a client and a cluster. We anchor the unidirectional half-pipe from the client to the cluster at a designated server while allowing the half-pipe from the cluster to the client to migrate on a per-request basis to an optimal server where the request is best serviced We describe the design and implementation of a multiple connection handoff mechanism in the Linux kernel that demonstrates the benefits of our technique.