In the past few years, video streaming over cellular networks has become increasingly popular due to which managing end-user Quality of Experience (QoE) has become a crucial problem. We consider the problem of variable bit rate (VBR) video streaming over heterogeneous cellular networks (HetNets). We address this problem as a joint assignment and scheduling problem and design QoE-aware greedy algorithms to solve it. The greedy assignment algorithm links users to femtocells for downloading the videos from, and the greedy scheduling algorithm fairly allocates time slots to the users assigned to a specific femtocell. Our experiments on real-world 4G LTE wireless connectivity and VBR video traces show the superior performance of our algorithm.

Accurate wireless channel quality prediction over 4G LTE networks continues to be an important problem as future channel predictions are widely leveraged to meet the strict requirements of applications such as 360-degree video, ARlVR, and online games. The availability of large amounts of wireless channel data, the increase in computational power and the advancements in the field of machine learning provide us the opportunity to design learning-based approaches to address the channel quality prediction problem. In this paper, we design discriminative sequence-to-sequence probabilistic graphical models, specifically sparse Gaussian Conditional Random Fields (GCRF) models to accurately predict future channel quality variations in 4G LTE networks based on past channel quality data. In contrast to prior work that has primarily focused on designing parsimonious Markovian models or computationally-intensive deep learning models, the sparse GCRF models designed here provide superior performance while being highly interpretable and computationally efficient, thus making them an ideal choice for practical deployment. To validate the efficacy of our sparse GCRF model, we compare its performance (i.e., root mean squared error and mean absolute error) with i) linear regression and ii) ARIMA and iii) the state-of-the-art deep learning model on real-world 4G LTE channel quality data collected under varying levels of user mobility for two cellular operators and observe that the GCRF model provides significantly higher performance improvement.

In this paper, our goal is to model the aggregate mobility of individuals in a city by analyzing cellular network connections, and then leverage the designed mobility model to model and predict the number of COVID-19 infections in future. We analyze cellular network connections from 973 antennas for all users in the city of Rio de Janeiro from April 5, 2020 to July 2, 2020. We design a Markovian model that captures the mobility across municipalities. We then combine the transition probabilities of the Markov chain with the number of COVID-19 cases in a municipality during a particular week in the design of our mobility-aware COVID-19 case prediction models to predict the number of cases for the following week. Our experiments demonstrate that our mobility-aware models significantly out-perform a baseline mobility-agnostic linear regression model in terms of metrics such as Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE).

In this paper, our goal is to analyze and compare cellular network usage data from pre-lockdown, during lock-down, and post-lockdown phases surrounding the COVID-19 pandemic to understand and model human mobility patterns during the pandemic. To this end, we collect and analyze cellular network connections from 1400 antennas for all users in the city of Rio de Janeiro and its suburbs from March 1, 2020 to July 1, 2020. Our analysis reveals that the total number of cellular connections decreases to 78% during the lockdown phase and then increases to 85% of the pre-COVID era as the lockdown eases. We observe that user mobility starts increasing around 3 weeks before the end of lockdown, with the trend continuing into the post-lockdown period. We also design an interactive tool that showcases mobility patterns in different granularities and can help government officials take informed actions to control the spread of the disease.

In this paper, our goal is to leverage cellular network traffic data to model and forecast the number of COVID-19 infections in the future. To this end, we partner with one of the main cellular network providers in Brazil, TIM Brazil, and collect and analyze cellular network connections from 973 antennas for all users in the city of Rio de Janeiro and its suburbs. We develop a Markovian model that captures the mobility of individuals across municipalities of the city. The transition probabilities of the Markov chain are determined by analyzing user-level mobility events between antennas from the cellular network connectivity logs. We combine the aggregate mobility characteristics across municipalities as evidenced from the transition probabilities with the number of reported COVID-19 cases in a municipality during a particular week to design mobility-aware COVID-19 case prediction models that predict the number of cases for the following week. Our experiments demonstrate that our mobility-aware models significantly outperform a baseline mobility-agnostic linear regression model in terms of metrics such as Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE).

In this paper, our goal is to analyze and compare cellular network usage data from Rio de Janeiro from pre-lockdown, during lockdown, and post-lockdown phases surrounding the COVID-19 pandemic to understand and model human mobility patterns during the pandemic, and to evaluate the effect of lockdowns on mobility. Our analysis reveals that human mobility increases significantly even before lockdown restrictions are eased, with the trend continuing in the post-lockdown period. We also observe that the day of week has a significant impact on mobility of individuals, with the overall mobility on Fridays increasing over time possibly due to people self-relaxing restrictions and engaging in social activities on Friday evenings. We also design an interactive tool that showcases mobility patterns in different granularities and can potentially help people and government officials understand the mobility of individuals and the number of COVID-19 cases in a particular neighborhood.

In this paper, we collect data from Twitter and conduct a linguistic analysis of the user tweets to understand the social and economic disruption caused by the COVID-19 pandemic. To better appreciate peoples’ opinions and concerns with regards to the socio-economic conditions of addiction, mental health, unemployment and immigration, we collect data for a period of approximately 3 months in the beginning of the pandemic. We analyze the term and co-occurrence frequencies to identify the most commonly occurring words and bigrams in the discussion for each of the four categories. We conduct semantic role labeling to determine the action words in each category and then adopt a LSTM-based dependency parsing model to identify the main nouns linked with these action words. We then adopt a seeded topic modeling approach to automatically identify the main topics of discussion in each category. We finally conclude with a sentiment analysis of the tweets in each category to determine the overall sentiment associated with each category. Our fine-grained linguistic study unearths the difficulties experienced by the people (e.g., action verb need associated with nouns such as aid and assistance in the unemployment category). We also observe that the overall sentiment in the tweets is negative, driven by people experiencing the pains of job loss, deportation, and the difficulty in accessing programs and treatments related to addiction. Our analysis highlights the main challenges experienced by the people during the start of the COVID-19 crisis and lays the foundation for recognizing and developing the most pertinent public and social policies so as to minimize peoples’ suffering in case of a future pandemic.

Accurately modeling and predicting wireless channel quality variations is essential for a number of networking applications such as scheduling and improved video streaming over 4G LTE networks and bit rate adaptation for improved performance in WiFi networks. In this paper, we propose an encoder-decoder based sequence-to-sequence deep learning model that is capable of predicting future wireless signal strength variations based on past signal strength data. We consider two different versions of the deep learning model; the first and second versions use LSTM and GRU as their basic cell structure, respectively. In contrast to prior work that is primarily focused on designing models for particular network settings, the deep learning model is highly adaptable and can predict future channel conditions for different networks, sampling rates, mobility patterns, and communication standards. We compare the performance (i.e., the root mean squared error of future predictions) of our model with respect to two baselines-i) auto-regression(1), and ii) linear regression for multiple networks and communication standards. In particular, we consider 4G LTE, WiFi, an industrial network operating in the 5.8 GHz range, Zigbee, and WiMAX networks operating under varying levels of user mobility and observe that the deep learning model provides significantly superior performance. Finally, we provide detailed discussion on key design decisions including insights into hyper-parameter tuning of the model.

In traditional mobile networks that do not support in-network caching, node mobility can cause a large number of requests to get dropped due to unavailability of paths from users to the origin servers (content custodians). In comparison, in information-centric networks (ICN) that cache content at storage-enabled network nodes in addition to the content custodians, a significantly higher percentage of requests can be served by effectively leveraging in-network caching even when direct paths are unavailable. In this paper, we propose QuickR, a routing algorithm that augments shortest path routing with features such as multi-path routing, content search, random walk, request caching and path recomputation to provide superior performance in wireless mobile ICN. We propose two versions of QuickR-i) a low-overhead version 1 that includes multi-path routing, neighbor search, random walk and request caching, and ii) version 2 that includes route recomputation along with all the other features of version 1. To demonstrate its robustness, we evaluate the performance of QuickR on diverse mobile networks-synthetic mobility models (i.e., Grid, Random waypoint), pedestrian mobility traces (i.e., Stockholm pedestrian trace) and real-world vehicular mobility traces (i.e., Rome taxi cab, Seattle bus). We also test QuickR's performance on real-world request streams from YouTube and Wikipedia, and on multiple cache eviction and insertion policies. Our experiments demonstrate that QuickR outperforms shortest path and multi-path routing in terms of percentage of requests served by a factor of 3. 5x on average.

In today's busy modern life, modeling and accurately predicting fitness center equipment usage and availability is essential for improving human fitness and well-being as it provides people the flexibility to plan their schedule and exercise at their convenience. In addition to its crucial role in ensuring a healthy and sustainable future, adopting a data-driven approach for modeling and predicting fitness center equipment usage is necessary for planning the optimal square footage for developing a fitness center, and determining the kinds of equipment to purchase and install. In this paper, we develop DeepFit, a deep learning based system that predicts future fitness center equipment usage based on historical data. To this end, we design a Long Short Term Memory (LSTM) based sequence-to-sequence model that captures the dependencies in the data. The sequence-to-sequence model comprises of an encoder and a decoder, each of which separately is a deep Recurrent Neural Network (RNN). The basic cell structure in the RNN architecture is an LSTM cell.

Caching content at storage-enabled network nodes is one of the salient features of Information-centric Networking (ICN). Most existing ICN caching and routing strategies have been designed for and evaluated in static networks [2], while mobile networks have received limited attention [1]. In this paper, we consider a mobile ICN and examine the benefits that in-network caching can provide in a mobile setting. From our experiments on mobile grid networks, we observe that approximately 43% of requests cannot be routed and hence are dropped in a mobile network due to absence of direct paths from users to the content custodian (origin server). For the requests served, we observe that while state-of-the-art caching strategies outperform the baseline Leave Copy Everywhere (LCE) strategy in static networks, all caching strategies including LCE tend to provide similar performance as network mobility increases. For requests that cannot be routed due to unavailability of paths from users to the custodian, we next explore the benefits of augmenting shortest path routing with a simple stateless random walk request forwarding strategy that enables a user to search neighboring caches. We observe that just using a simple random walk strategy when a direct path is unavailable increases the total percentage of requests served by 6% on average.

To address the explosive increase in mobile data traffic in recent years, content caching at storage-enabled network nodes has been proposed. Alongside, a variety of forwarding strategies have been developed for wireless networks that exploit the broadcast nature of the wireless medium and the presence of time-varying fading channels to improve user performance. A widely popular forwarding strategy is opportunistic routing that opportunistically selects nodes that overhear packet transmissions to serve as ad hoc relays to forward the packet. In this paper, we investigate the request routing delay of a greedy opportunistic routing strategy for cache-enabled wireless networks considering uncorrelated and temporally correlated Rayleigh fading wireless channels. To this end, we develop Markovian models, leverage the wireless channel characteristics to determine the transition probabilities and then utilize them to obtain the request routing delay. Via numerical evaluation and simulation, we demonstrate the validity and effectiveness of our model in determining the request routing delay. We also investigate the impact of network parameters on performance through our experiments. Our work takes a step forward in providing network operators a tool for analyzing network performance before deploying their networks.

With the increasing popularity of cache networks, in recent years, multiple static and dynamic caching strategies have been proposed that seek to improve user-level performance. Most existing caching strategies rely heavily on assumptions such as content popularity following a well-known Zipfian distribution and request streams following an Independent Reference Model (IRM). In this paper, we consider multiple real-world user request stream traces to investigate the validity of these assumptions and observe that they do not hold true. We conduct a detailed factor analysis and observe that violation of the IRM assumption significantly impacts the performance of caching strategies. We identify the interplay between the skewness of the content popularity distribution and the request stream correlation among unpopular content as the key factor impacting performance. We identify that in the high popularity skewness-low correlation regime, static caching outperforms dynamic caching, while the reverse is true in the low popularity skewness-high correlation regime. For the high popularity skewness-high correlation regime, static and dynamic caching provide similar performance. For this scenario, we propose Hybrid Caching that effectively combines static and dynamic caching strategies. The main idea is to split the cache into two parts-a static cache that statically caches content based on popularity and a dynamic cache that exploits the correlation in request streams. We conduct experiments on multiple real-world networks (e.g., WIDE, GEANT, GARR) and demonstrate that Hybrid Caching outperforms static or dynamic caching alone in the high popularity skewness-high correlation regime.

Cities have limited resources that must be used efficiently to maintain their smooth operation. To facilitate efficient resource allocation and management in cities, in this paper, we study one such important problem: how long does it take to resolve non-emergency 311 service requests? We present NYCER, a Non-emergency Response prediction system based on a recently developed structured regression model, sparse Gaussian conditional random fields (GCRFs), that successfully captures the dependencies between historical and future response times. Through extensive experimentation on 311 service requests in New York City (NYC) over a three and a half year period between Jan 2015 to June 2018, we demonstrate that our trained system is able to accurately predict future response times one week in advance using just the previous two weeks data at test time. NYCER achieves superior prediction performance across all agencies, complaint types, and locations, when compared to a linear regression baseline (up to a factor of 2X). The trained NYCER system requires low computational resources and data at test time, thus making it an attractive system that can be readily deployed in practice.

Water availability and management is an important problem plaguing many developing and under-developed countries. Many factors including geographic, political, management, and environmental factors affect the availability of water in these regions. In this paper, we develop an ensemble-learning based predictive-analytics framework for smart water management to predict: i) water pump operation status (e.g., functional, non functional), ii) water quality, and iii) quantity. In the predictive-analytics framework, we first perform feature engineering to select relevant features, use them to develop the XGBoost and Random Forest ensemble learning models, and then perform extensive feature analysis to identify the most predictive features, for each prediction problem mentioned above. We evaluate our framework on two publicly available smart water management datasets pertaining to Tanzania and Nigeria and show that our proposed models outperform several baseline approaches, including logistic regression, SVMs, and multi-layer perceptrons in terms of precision, recall and F1 score. We also demonstrate that our models are able to achieve a superior prediction performance for predicting water pump operation status for different water extraction methods. We conduct a detailed feature analysis to investigate the importance of the various feature groups (e.g., geographic, management) on the performance of the models for predicting water pump operation status, water quality and quantity. We then perform a fine-grained feature analysis to identify how individual features, not just feature groups, impact performance. We identify that among individual features, location (x, y, z coordinates) has the maximum impact on performance. Our analysis is helpful in understanding the types of data that should be collected in future for accurately predicting the different water problems.

Reducing the carbon footprint of energy generation is an important part of ongoing sustainability efforts. To cut carbon footprints, electric utilities are incentivizing renewable energy integration through net metering and introducing time-of-use pricing plans to cut demand peaks, as peaks significantly contribute to both generation costs and carbon emissions. Net metering is one of the most popular means of integrating distributed renewable generation in the grid. However, the current net metering approach doesn't effectively cut demand peaks because renewable harvest peak and demand peaks are out of sync. Furthermore, as several states impose net metering subscriber limits of less than 1% of the peak, net metering isn't even close to realizing the full potential of renewable integration in the grid. To address these limitations, we present GreenPeaks, an energy storage based renewable integration system to enhance net metering. GreenPeaks employs energy storage to intelligently move a fraction of harvested energy to peak intervals and accumulate any surplus harvest. We evaluate GreenPeaks using consumption data from real homes. Our results show that GreenPeaks reduces grid-wide peak by 12% in contrast to net metering's 2%, while reducing the electricity generation costs by more than 40%.

With the growth of Internet-of-Things, mobile data traffic is expected to increase exponentially. To support this rapid growth, heterogeneous cellular networks comprising of femtocells with storage capabilities along with macrocell base stations have been proposed. In this paper, we first investigate the performance impact of a simple randomized cache pollution attack, where the attacker pollutes the cache at the femtocell by requesting unpopular content. We then adopt a principled approach based on the characteristic time of a content in a cache to design an optimized attack strategy. Our experiments show that the proposed attack strategy outperforms the randomized attack with the same attack rate.

In-network content caching has been deployed in both the Internet and cellular networks to reduce content-access delay. We investigate the problem of developing optimal joint routing and caching policies in a network supporting in-network caching with the goal of minimizing expected content-access delay. Here, needed content can either be accessed directly from a back-end server (where content resides permanently) or be obtained from one of multiple in-network caches. To access content, users must thus decide whether to route their requests to a cache or to the back-end server. In addition, caches must decide which content to cache. We investigate two variants of the problem, where the paths to the back-end server can be considered as either congestion-sensitive or congestion-insensitive, reflecting whether or not the delay experienced by a request sent to the back-end server depends on the request load, respectively. We show that the problem of optimal joint caching and routing is NP-complete in both cases. We prove that under the congestion-insensitive delay model, the problem can be solved optimally in polynomial time if each piece of content is requested by only one user, or when there are at most two caches in the network. We also identify the structural property of the user-cache graph that makes the problem NP-complete. For the congestion-sensitive delay model, we prove that the problem remains NP-complete even if there is only one cache in the network and each content is requested by only one user. We show that approximate solutions can be found for both cases within a $(1-1/e)$ factor from the optimal, and demonstrate a greedy solution that is numerically shown to be within 1% of optimal for small problem sizes. Through trace-driven simulations, we evaluate the performance of our greedy solutions to joint caching and routing, which show up to 50% reduction in average delay over the solution of optimized routing to least recently used caches.

We investigate the problem of optimal request routing and content caching in a heterogeneous network supporting in-network content caching with the goal of minimizing average content access delay. Here, content can either be accessed directly from a back-end server (where content resides permanently) or be obtained from one of multiple in-network caches. To access a piece of content, a user must decide whether to route its request to a cache or to the back-end server. Additionally, caches must decide which content to cache. We investigate the problem complexity of two problem formulations, where the direct path to the back-end server is modeled as i) a congestion-sensitive or ii) a congestion-insensitive path, reflecting whether or not the delay of the uncached path to the back-end server depends on the user request load, respectively. We show that the problem is NP-complete in both cases. We prove that under the congestion-insensitive model the problem can be solved optimally in polynomial time if each piece of content is requested by only one user, or when there are at most two caches in the network. We also identify a structural property of the user-cache graph that potentially makes the problem NP-complete. For the congestion-sensitive model, we prove that the problem remains NP-complete even if there is only one cache in the network and each content is requested by only one user. We show that approximate solutions can be found for both models within a (1 - 1/e) factor of the optimal solution, and demonstrate a greedy algorithm that is found to be within 1% of optimal for small problem sizes. Through trace-driven simulations we evaluate the performance of our greedy algorithms, which show up to a 50% reduction in average delay over solutions based on LRU content caching.

A variety of forwarding strategies have been developed for multi-hop wireless networks, considering the broadcast nature of the wireless medium and the presence of fading channels that result in time-varying and unreliable transmission quality. One such strategy is opportunistic forwarding, which exploits relay diversity by opportunistically selecting an overhearing relay as a forwarder. Prior work has studied the performance of opportunistic forwarding for the simplified scenario of uncorrelated wireless channels. In this paper, we consider a more realistic scenario of temporally correlated wireless channels; the wireless channel is modeled as a Rayleigh fading channel and its temporal correlation as a modified Bessel function of the first kind and zeroth order. We use these models to develop a simple Markovian model to analyze the performance of opportunistic forwarding for correlated wireless channels for the case of linear networks. We then demonstrate via numerical evaluation the diminishing performance of opportunistic forwarding with increasing channel correlation.

Managing the Quality-of-Experience (QoE) of video streaming for wireless clients is becoming increasingly important due to the rapid growth of video traffic on wireless networks. The inherent variability of the wireless channel as well as the Variable Bit Rate (VBR) of the compressed video streams make QoE management a challenging problem. In this paper, we investigate scheduling algorithms to transmit multiple video streams from a base station to mobile clients. We present an epoch-by-epoch framework to fairly allocate wireless transmission slots to streaming videos. In each epoch, our scheme reduces the vulnerability to stalling by allocating slots to videos in a way that maximizes the minimum “playout lead” across all videos. We show that the problem of allocating slots fairly is NP-complete even for a constant number of videos. We then present a fast lead-aware greedy scheduling algorithm. Our greedy algorithm is optimal when the channel quality of a user remains unchanged within an epoch. Our experimental results, based on public MPEG-4 video traces and wireless channel traces that we collected from a WiMAX test-bed, show that the lead-aware greedy approach results in a fair distribution of stalls across the clients when compared to other algorithms, while still maintaining similar or fewer average number of stalls per client.

Managing the Quality-of-Experience (QoE) of video streaming for wireless clients is becoming increasingly important due to the rapid growth of video traffic on wireless networks. The inherent variability of the wireless channel as well as the Variable Bit Rate (VBR) of the compressed video streams make QoE management a challenging problem. Prior work has studied this problem in the context of transmitting a single video stream. In this paper, we investigate multiplexing schemes to transmit multiple video streams from a base station to mobile clients that use number of playout stalls as a performance metric. In this context, we present an epoch-by-epoch framework to fairly allocate wireless transmission slots to streaming videos. In each epoch our scheme essentially reduces the vulnerability to stalling by allocating slots to videos in a way that maximizes the minimum `playout lead' across all videos. Next, we show that the problem of allocating slots fairly is NP-complete even for a constant number of videos. We then present a fast lead-aware greedy algorithm for the problem. Our choice of greedy algorithm is motivated by the fact that this algorithm is optimal when the channel quality of a user remains unchanged within an epoch (but different users may experience different channel quality). Moreover, our experimental results based on public MPEG-4 video traces and wireless channel traces that we collected from a WiMAX test-bed show that the lead-aware greedy approach performs a fair distribution of stalls across the clients when compared to other algorithms, while still maintaining similar or lower average number of stalls per client.

A wide range of wireless channel models have been developed to model variations in received signal strength. In contrast to prior work, which has focused primarily on channel modeling on a short, per- packet timescale (millisecond), we develop and validate a finite-state Markov chain model that captures variations due to shadowing, which occur at coarser time scales. The Markov chain is constructed by partitioning the entire range of shadowing into a finite number of intervals. We determine the Markov chain transition matrix in two ways: (i) via an abstract modeling approach in which shadowing effects are modeled as a log-normally distributed random variable affecting the received power, and the transition probabilities are derived as functions of the variance and autocorrelation function of shadowing; (ii) via an empirical approach, in which the transition matrix is calculated by directly measuring the changes in signal strengths collected in a 802.16e (WiMAX) network. We validate the abstract model by comparing its steady state and transient performance predictions with those computed using the empirically derived transition matrix and those observed in the actual traces themselves.

We present BlockRate, a wireless bitrate adaptation algorithm designed for blocks, or large contiguous units of transmitted data, as opposed to small packets. Our work is motivated by the observation that recent research results suggest significant overhead amortization benefits of blocks. Yet state-of-the-art bitrate algorithms are optimized for adaptation on a per-packet basis, so they can either have the amortization benefits of blocks or high responsiveness to underlying channel conditions of packets, but not both.