The ubiquitous deployment of 4G/5G technology has made it a critical infrastructure for society that will facilitate the delivery and adoption of emerging applications and use cases (extended reality, automation, robotics, to name but a few). These new applications require high throughput and low latency in both uplink and downlink for optimal performance, while coexisting with traditional downlink-heavy consumer applications. Successfully supporting these new use cases hinges on the network being able to allocate resources as efficiently as possible. In this paper, we utilize a 3GPP-compliant 5G testbed to analyze the limitations of legacy network resource allocation methods, which are based on instantaneous channel measurements, and examine the effect on throughput – a key performance indicator. We then propose a framework that allows resource allocation decisions to leverage predictions of network quality (computed at the connected devices), and study two different prediction methods that provide different degrees of reliability. We further validate our framework with real-world cellular data and demonstrate that with accurate channel metric forecast knowledge, the mean network throughput can improve by a factor of $\sim 1.8$ over the baseline reactive approach based on best CQI policy, for the considered scenario.

Recent advancements in deep learning techniques have transformed the area of semantic text matching (STM). However, most state-of-the-art models are designed to operate with short documents such as tweets, user reviews, comments, and so on. These models have fundamental limitations when applied to long-form documents such as scientific papers, legal documents, and patents. When handling such long documents, there are three primary challenges: (i) the presence of different contexts for the same word throughout the document, (ii) small sections of contextually similar text between two documents, but dissimilar text in the remaining parts (this defies the basic understanding of “similarity”), and (iii) the coarse nature of a single global similarity measure which fails to capture the heterogeneity of the document content. In this article, we describe CoLDE : Co ntrastive L ong D ocument E ncoder—a transformer-based framework that addresses these challenges and allows for interpretable comparisons of long documents. CoLDE uses unique positional embeddings and a multi-headed chunkwise attention layer in conjunction with a supervised contrastive learning framework to capture similarity at three different levels: (i) high-level similarity scores between a pair of documents, (ii) similarity scores between different sections within and across documents, and (iii) similarity scores between different chunks in the same document and across other documents. These fine-grained similarity scores aid in better interpretability. We evaluate CoLDE on three long document datasets namely, ACL Anthology publications, Wikipedia articles, and USPTO patents. Besides outperforming the state-of-the-art methods on the document matching task, CoLDE is also robust to changes in document length and text perturbations and provides interpretable results. The code for the proposed model is publicly available at https://github.com/InterDigitalInc/CoLDE .

In this paper, we analyze a large dataset of passive wireless measurements and obtain insights about wireless performance. We monitor 167 homes continuously for 4 months from the vantage point of the gateway, which allows us to capture all the activity on the home wireless network. We report on the makeup of the home wireless network, traffic activity, and performance characteristics. We find that in most homes, a small number of devices account for most of the observed traffic volume and the bulk of this traffic activity occurs in the evenings. Studying link performance, we find that overall, the vast majority of transmissions are carried out at high data rates and the wireless networks have good coverage. We find a small number of episodes where performance is poor; a few homes have a disproportionate number of poor performance reports. Investigating further, we observe that most of these are not caused by poor coverage (pointing to network interference). Our results significantly add to the understanding of home wireless networks and will help ISPs to understand their subscriber networks.

In recent years, ISPs in mature residential broadband markets have been moving away from flat-rate pricing schemes to ones that involve usage limits (or "data caps"). These changes have led to a lively debate about the merits of data capping. While previous work has studied the theoretical benefits of data capped pricing models, there is relatively little work that provides insights into exactly how to design data capped tariffs, and their impact on the parties involved, i.e., ISPs and their broadband subscribers. In this paper, we formulate the problem of selecting optimal data capped tariffs based on a simple, extensible model of ISP costs and revenues, and a novel model that captures how users evaluate a set of offered tariffs.

Modern mobile devices are equipped with multiple network interfaces, with WiFi to be the preferred choice when it can be associated with an Access Point. Our study using real traces and experiments with 802.11 devices uncovers long idle times, while WiFi is scanning for networks and is listening for incoming traffic. During idle, WiFi consumes a significant amount of energy, which is exacerbated with the 802.11n/ac high speed features (MIMO, channel-bonding). State of the art power saving solutions, including the IEEE 802.11 PSM and SMPS features are user-agnostic, infrastructure-driven and oblivious to the power hungry 802.11n/ac features. As a result they may not save energy in practical settings. To this end, this work presents WiFi idle energy save (WiFi-ies). WiFi-ies leverages user context and behavior information to identify idle times where energy can be saved. We verify WiFi-ies energy savings using trace-driven simulations and prototype implementations on Android smartphones.

To address the pressing need to provide transparency into the online targeted advertising ecosystem, we present AdReveal, a practical measurement and analysis framework, that provides a first look at the prevalence of different ad targeting mechanisms. We design and implement a browser based tool that provides detailed measurements of online display ads, and develop analysis techniques to characterize the contextual, behavioral and re-marketing based targeting mechanisms used by advertisers. Our analysis is based on a large dataset consisting of measurements from 103K webpages and 139K display ads. Our results show that advertisers frequently target users based on their online interests; almost half of the ad categories employ behavioral targeting. Ads related to Insurance, Real Estate and Travel and Tourism make extensive use of behavioral targeting. Furthermore, up to 65% of ad categories received by users are behaviorally targeted. Finally, our analysis of re-marketing shows that it is adopted by a wide range of websites and the most commonly targeted re-marketing based ads are from the Travel and Tourism and Shopping categories.

We model a little studied type of traffic, namely the network traffic generated from endhosts. We introduce a parsimonious model of the marginal distribution for connection arrivals consisting of mixture models with both heavy and light-tailed component distributions. Our methodology assumes that the underlying user data can be fitted to one of several models, and we apply Bayesian model selection criterion to choose the preferred combination of components. Our experiments show that a simple Pareto-exponential mixture model is preferred over more complex alternatives, for a wide range of users. This model has the desirable property of modeling the entire distribution, effectively clustering the traffic into the heavy-tailed as well as the non-heavy-tailed components. Also this method quantifies the wide diversity in the observed endhost traffic.

We design predictors of user dissatisfaction with the performance of applications that use networking. Our approach combines user-level feedback with low level machine and networking metrics. The main challenges of predicting user dissatisfaction, that arises when networking conditions adversely affect applications, comes from the scarcity of user feedback and the fact that poor performance episodes are rare. We develop a methodology to handle these challenges. Our method processes low level data via quantization and feature selection steps. We combine this with user labels and employ supervised learning techniques to build predictors. Using data from 19 personal machines, we show how to build training sets and demonstrate that non-linear SVMs achieve higher true positive rates (around 0.9) than predictors based on linear models. Finally we quantify the benefits of building per-application predictors as compared to general predictors that use data from multiple applications simultaneously to anticipate user dissatisfaction.

Users today connect to the Internet everywhere - from home, work, airports, friend's homes, and more. This paper characterizes how the performance of networked applications varies across networking environments. Using data from a few dozen end-hosts, we compare the distributions of RTTs and download rates across pairs of environments. We illustrate that for most users the performance difference is statistically significant. We contrast the influence of the application mix and environmental factors on these performance differences.

Enterprise and data center networks consist of a large number of complex networked applications and services that depend upon each other. For this reason, they are difficult to manage and diagnose. In this paper we propose Macroscope, a new approach to extracting the dependencies of networked applications automatically by combining application process information with network level packet traces. We evaluate Macroscope on traces collected at 52 laptops within a large enterprise and show that Macroscope is accurate in finding the dependencies of networked applications. We also show that Macroscope requires less human involvement and is significantly more accurate than state of the art approaches that use only packet traces. Using our rich profiles of the application-service dependencies, we explore and uncover some interesting characteristics about this relationship. Finally, we discuss several usage scenarios that can benefit from Macroscope.

In this paper we take a look at the health of a typical enterprise network via a new metric based on the fraction of useful flows generated by endhosts. Flows considered non-useful are those that explicitly fail or else do not elicit a response from the intended destination. Examining traces collected from a large number of mobile hosts in an enterprise network, we find that about 34% of the flows are not useful. Through our study that combines data analysis and ongoing interactions with our IT department, we learn that these non-useful flows arise from several causes. Our mobile hosts frequently change environments, by either moving in and out of the corporate environment, or by switching the point and means of attachment to the corporate network. We find that many of the failures occur due to the hosts' lack of environment awareness, which results in attempts to discover services that are not present in all environments. Other causes include misconfiguration, unnecessary broadcast traffic, and excessive connection retries. Understanding this ever present noise in endhost communication is important for a variety of reasons including the fact that it complicates anomaly detection design and wastes resources, the latter of which is particularly crucial for wireless and mobile environments. Finally, we discuss possible means to design applications and services that can significantly improve the health of the network.

The distributed denial-of-service (DDoS) attack is a serious threat to the legitimate use of the Internet. Prevention mechanisms are thwarted by the ability of attackers to forge or spoof the source addresses in IP packets. By employing IP spoofing, attackers can evade detection and put a substantial burden on the destination network for policing attack packets. In this paper, we propose an interdomain packet filter (IDPF) architecture that can mitigate the level of IP spoofing on the Internet. A key feature of our scheme is that it does not require global routing information. IDPFs are constructed from the information implicit in border gateway protocol (BGP) route updates and are deployed in network border routers. We establish the conditions under which the IDPF framework correctly works in that it does not discard packets with valid source addresses. Based on extensive simulation studies, we show that, even with partial deployment on the Internet, IDPFs can proactively limit the spoofing capability of attackers. In addition, they can help localize the origin of an attack packet to a small number of candidate networks.

a serious threat to the legitimate use of the Internet. Prevention mechanisms are thwarted by the ability of attackers to forge, or spoof, the source addresses in IP packets. By employing IP spoofing, attackers can evade detection and put a substantial burden on the destination network for policing attack packets. In this paper we propose an inter-domain packet filter (IDPF) architecture that can mitigate the level of IP spoofing on the Internet. IDPFs are constructed from the information implicit in BGP route updates and are deployed in network border routers. A key feature of the scheme is that it does not require global routing information. Based on extensive simulation studies, we show that even with partial deployment on the Internet, IDPFs can proactively limit the spoofing capability of attackers. In addition, they can help localize the origin of an attack packet to a small number of candidate networks. I.

Slow convergence in the Internet can be directly attributed to the "path exploration" phenomenon, inherent in all path vector protocols. The root cause for path exploration is the dependency among paths propagated through the network. Addressing this problem in BGP is particularly difficult as the AS paths exchanged between BGP routers are highly summarized. In this paper, we describe why path exploration cannot be countered effectively within the existing BGP framework, and propose a simple, novel mechanism - forward edge sequence numbers - to annotate the AS paths with additional "path dependency" information. We then develop an enhanced path vector algorithm, EPIC, shown to limit path exploration and lead to faster convergence. In contrast to other solutions, ours is shown to be correct on a very general model of Internet topology and BGP operation. Using theoretical analysis and simulations, we demonstrate that EPIC can achieve a dramatic improvement in routing convergence, compared to BGP and other existing solutions.

Route flap damping (RFD) is anecdotally considered to be a key contributor in the stability of the inter-domain routing system. It works by suppressing advertisements about persistently flapping routes, which otherwise would propagate throughout the Internet. It was recently shown that relatively stable routes, i.e., routes that fail occasionally, can be incorrectly suppressed by this mechanism for substantially long periods of time. This can be traced back to the complex interaction between BGP path exploration and the mechanism used by RFD to identify route flaps. In this paper we study the distinctive feature that distinguishes the sequence of updates following a single network event from that of persistently unstable routes. Based on this characteristic, we propose a new BGP route flap damping algorithm, RFD+, with the following properties - 1) it can correctly distinguish between route flaps and normal path exploration; 2) it suppresses routes that are frequently and persistently changing; and 3) it does not affect routes that fail occasionally. We present the algorithm and discuss its relevant properties; simulation studies are also conducted to illustrate the performance of our algorithm.

Effective service delivery capabilities are critical to the transformation of the Internet into a viable commercial infrastructure. At the same time, there are several design limitations that prevent this. We propose a novel service overlay architecture that serves as a flexible, unifying platform for delivering services over the Internet. We introduce a new addressing scheme and an associated service layer, which enables service-oriented routing and forwarding over the underlying IP network domain. We also describe the functionality of the network elements that are introduced by our architecture, namely service gateway (SG) and service point-of-presence (S-PoP). We also present examples to demonstrate the efficacy of our architecture.