145 research outputs found
Lens : Leveraging social networking and trust to prevent spam transmission
No full textAbstract—In this paper we introduce LENS, a novel spam protection system based on the recipient’s social network, which allows correspondence within the social circle to directly pass to the mailbox and further mitigates spam beyond social circles. The key idea in LENS is to select legitimate and authentic users, called Gatekeepers (GKs), from outside the recipients social circle and within pre-defined social distances. Unless a GK vouches for the emails of potential senders from outside the so-cial circle of a particular recipient, those e-mails are prevented from transmission. In this way LENS drastically reduces the consumption of Internet bandwidth by spam. Using extensive evaluations, we show that LENS provides each recipient reliable email delivery from a large fraction of the social network. We also evaluate the computational complexity of email processing with LENS deployed on two Mail Servers (MSs) and compared it with the most popular content-based filter i.e SpamAssassin. LENS proved to be fast in processing emails (around 2-3 orders of magnitude better than SpamAssassin) and scales efficiently with increasing community size and GKs. I
Fighting Spam Using Social GateKeepers
No full textWe introduce LENS (LEveraging social Networking and trust to prevent Spam transmission), a novel spam protection system which leverages the recipient’s social network to allow correspondence within the social network to directly pass to the mailbox of the recipient. To enable new senders to send emails, legitimate and authentic users, called GateKeepers (GKs), are selected from outside the recipient’s social circle and within predefined social distances. Our evaluations show that LENS provides each recipient reliable email delivery from a large fraction (up to 55% of entire userbase) of the social network; it is also effective and lightweight in accepting all the legitimate inbound emails in the real email traces. LENS imposes zero overhead for the common case of frequent and familiar senders, and remains lightweight for the general case. Our prototype implementation of LENS in Postfix/MailAvenger shows that LENS consumes up to 75% less CPU and 9% less memory as traditional solutions like SpamAssassin
FaceLift: a transparent deep learning framework to beautify urban scenes
No full textIn the area of computer vision, deep learning techniques have recently been used to predict whether urban scenes are likely to be considered beautiful: it turns out that these techniques are able to make accurate predictions. Yet they fall short when it comes to generating actionable insights for urban design. To support urban interventions, one needs to go beyond predicting beauty, and tackle the challenge of recreating beauty. Unfortunately, deep learning techniques have not been designed with that challenge in mind. Given their 'black-box nature', these models cannot be directly used to explain why a particular urban scene is deemed to be beautiful. To partly fix that, we propose a deep learning framework (which we name FaceLift1) that is able to both beautify existing urban scenes (Google Street Views) and explain which urban elements make those transformed scenes beautiful. To quantitatively evaluate our framework, we cannot resort to any existing metric (as the research problem at hand has never been tackled before) and need to formulate new ones. These new metrics should ideally capture the presence (or absence) of elements that make urban spaces great. Upon a review of the urban planning literature, we identify five main metrics: walkability, green spaces, openness, landmarks and visual complexity. We find that, across all the five metrics, the beautified scenes meet the expectations set by the literature on what great spaces tend to be made of. This result is further confirmed by a 20-participant expert survey in which FaceLift has been found to be effective in promoting citizen participation. All this suggests that, in the future, as our framework's components are further researched and become better and more sophisticated, it is not hard to imagine technologies that will be able to accurately and efficiently support architects and planners in the design of the spaces we intuitively love
Preserving Privacy of Vulnerable Users across Heterogeneous Sensitive Sensor Data Streams using Smart Contracts
No full textThis paper is concerned with helping people who are vulnerable during important transitions in life, such as 'coming out' as LGBTQIA+, experiencing serious illness, undergoing relationship breakdown etc. Rich sensor streams derived from so-called 'smart' Internet of Things (IoT) devices can be highly beneficial, for example in ensuring the safety of such individuals during their sensitive life transitions, or in providing functionality that can mitigate some of the difficulties faced by them. However, the data that needs to be extracted to provide these benefits can itself be highly sensitive and needs to be processed with safeguards to protect privacy. We develop scenarios that highlight issues arising from having to merge data streams from multiple devices, including data governance issues that are relevant when the sensors are owned by multiple individuals. We propose a "Transition Guardian" architecture that leverages "Smart Experts" written as smart contracts operating on homomorphically encrypted sensor data streams to provide real-time protection without disclosing their sensitive information. We have also implemented a proof-of-concept on the Ethereum protocol to validate our proposed solution
Who Has the Last Word? Understanding How to Sample Online Discussions
No full textIn online debates, as in offline ones, individual utterances or arguments support or attack each other, leading to some subset of arguments (potentially from different sides of the debate) being considered more relevant than others. However, online conversations are much larger in scale than offline ones, with often hundreds of thousands of users weighing in, collaboratively forming large trees of comments by starting from an original post and replying to each other. In large discussions, readers are often forced to sample a subset of the arguments being put forth. Since such sampling is rarely done in a principled manner, users may not read all the relevant arguments to get a full picture of the debate from a sample. This article is interested in answering the question of how users should sample online conversations to selectively favour the currently justified or accepted positions in the debate. We apply techniques from argumentation theory and complex networks to build a model that predicts the probabilities of the normatively justified arguments given their location in idealised online discussions of comments and replies, which we represent as trees. Our model shows that the proportion of replies that are supportive, the distribution of the number of replies that comments receive, and the locations of comments that do not receive replies (i.e., the “leaves” of the reply tree) all determine the probability that a comment is a justified argument given its location. We show that when the distribution of the number of replies is homogeneous along the tree length, for acrimonious discussions (with more attacking comments than supportive ones), the distribution of justified arguments depends on the parity of the tree level, which is the distance from the root expressed as number of edges. In supportive discussions, which have more supportive comments than attacks, the probability of having justified comments increases as one moves away from the root. For discussion trees that have a non-homogeneous in-degree distribution, for supportive discussions we observe the same behaviour as before, while for acrimonious discussions we cannot observe the same parity-based distribution. This is verified with data obtained from the online debating platform Kialo. By predicting the locations of the justified arguments in reply trees, we can therefore suggest which arguments readers should sample, to grasp the currently accepted opinions in such discussions. Our models have important implications for the design of future online debating platforms
An empirical study of the transport layer performance and security in mobile networks
No full textEnabling ultra-low latency, and high-speed reliable connectivity, the latest generations of mobile networks are equipped to cater to diverse business and application requirements. Technologies such as software-defined networking (SDN), control and user plane separation (CUPS), deep-learning-based AI solutions along with newer networking protocols such as QUIC play critical roles in mobile networks to ensure state-of-the-art network performance.
This thesis thoroughly reviews the evolving security threat landscape of the modern mobile networks, empirically investigate the impacts of CUPS hijacking of a radio access network (RAN) slicing system on the overall network performances, and finally presents an experimental evaluation and characterization of QUIC performance over commercial 5G network.
In emerging mobile networks, CUPS plays a critical role in scaling the control-plane and user-plane functions independently and enables network virtualization through network slicing. However, a CUPS hijacking attack on a mobile network slicing system and the resulting network performance degradation are yet to be studied.
We investigate the consequences of CUPS hijacking of a RAN slicing system on the overall network performance. We quantify the impacts of CUPS hijacking by designing an Impact Factor metric I, prototype a real-world RAN slicing use case on an end-to-end mobile network test-bed, and systematically analyze the empirical results to reveal the impacts of CUPS hijacking on the network performance. We show a successful CUPS hijacking by a rogue slice owner in a RAN slicing system increases the RAN slice control-plane signaling delay above 2ms, the operational upper-bound of our system, to disrupt the control plane operations by injecting low rate DoS (LDoS) traffic in user-plane. The naive hijacking can degrade the throughput performances of the rogue slice as well as a co-located victim slice down to 0 Mbps. We further show that carefully crafted user-plane traffic by the attacker can regain 92\% of its original user-plane packet delivery success rate while other slices are under the denial of service.
On the other hand, the new transport layer protocol QUIC is being adopted by major internet application providers such as Google and Facebook signifying a paradigm shift from the de-facto transport layer protocol TCP. However, the interactions between 5G as a network data link layer for QUIC transport protocol to deliver modern web browsing, video streaming, and file downloading applications remain unexplored.
We conduct an in-depth study of 5G and QUIC interactions and their impact on application performances by collecting measurements taken in the wild involving commercial 5G networks and production-grade QUIC application servers. Results from the study reveal that end-user experience of accessing popular web services such as 4K video streaming on YouTube, file downloading from the Google cloud server, Google Search, and Maps on QUIC remain similar to the experience when accessed on TCP, over mobile networks
Session reports for SIGCOMM 2010
No full textThis document collects together reports of the sessions from the 2010 ACM SIGCOMM Conference, the annual conference of the ACM Special Interest Group on Data Communication (SIGCOMM) on the applications, technologies, architectures, and protocols for computer communication.</jats:p
DandeBot - An Autonomous Weeding Solution for Residential Lawns
No full textDandeBot – An Autonomous Weeding Solution for Residential Lawns
Author: Nishanth Rajkumar
This thesis presents the development and validation of DandeBot, an autonomous robotic system designed for comprehensive residential lawn maintenance. The robot addresses the need for efficient, eco-conscious, and low maintenance lawn care through a fully electric platform powered by an AI-driven software stack. Emphasizing safety, adaptability, and ease of use, the hardware was developed using CAD and Design for Manufacturing (DFM) principles, resulting in a modular and robust design. The integrated software stack combines localization, mapping, and path planning using odometry, visual odometry, and IMU data fusion to navigate dynamic outdoor environments. Task-specific algorithms were developed and validated for autonomous navigation, weed detection, and obstacle avoidance. Key hardware innovations include a modular gripper system for weed removal and adaptable attachments for multiple lawn care tasks. Field trials confirmed the robot’s capability to perform with high precision and reliability in varied lawn conditions, significantly reducing the need for human intervention. This work contributes to the growing field of service robotics by demonstrating how intelligent systems can automate routine household maintenance. The thesis concludes by outlining future research directions, including system scalability, enhanced multi-tasking capabilities, and integration with smart home networks
- …
