Abstract
We develop a simple model for the timely monitoring of correlated sources over a wireless network. Using this model, we study how to optimize weighted-sum average Age of Information (AoI) in the presence of correlation. First, we discuss how to find optimal stationary randomized policies and show that they are at-most a factor of two away from optimal policies in general. Then, we develop a Lyapunov drift-based max-weight policy that performs better than randomized policies in practice and show that it is also at-most a factor of two away from optimal. Next, we derive scaling results that show how AoI improves in large networks in the presence of correlation. We also show that for stationary randomized policies, the expression for average AoI is robust to the way in which the correlation structure is modeled. Finally, for the setting where correlation parameters are unknown and time-varying, we develop a heuristic policy that adapts its scheduling decisions by learning the correlation parameters in an online manner. We also provide numerical simulations to support our theoretical results.
Mass Exit Attacks on the Lightning Network
Authors: Anastasios Sidiropoulos, Cosimo Sguanci
Subjects: Cryptography and Security (cs.CR); Distributed, Parallel, and Cluster Computing (cs.DC)
Abstract
The Lightning Network (LN) has enjoyed rapid growth over recent years, and has become the most popular scaling solution for the Bitcoin blockchain. The security of the LN hinges on the ability of the nodes to close a channel by settling their balances, which requires confirming a transaction on the Bitcoin blockchain within a pre-agreed time period. This inherent timing restriction that the LN must satisfy, make it susceptible to attacks that seek to increase the congestion on the Bitcoin blockchain, thus preventing correct protocol execution. We study the susceptibility of the LN to \emph{mass exit} attacks, in the presence of a small coalition of adversarial nodes. This is a scenario where an adversary forces a large set of honest protocol participants to interact with the blockchain. We focus on two types of attacks: (i) The first is a \emph{zombie} attack, where a set of $k$ nodes become unresponsive with the goal to lock the funds of many channels for a period of time longer than what the LN protocol dictates. (ii) The second is a \emph{mass double-spend} attack, where a set of $k$ nodes attempt to steal funds by submitting many closing transactions that settle channels using expired protocol states; this causes many honest nodes to have to quickly respond by submitting invalidating transactions. We show via simulations that, under historically-plausible congestion conditions, with mild statistical assumptions on channel balances, both of the attacks can be performed by a very small coalition. To perform our simulations, we formulate the problem of finding a worst-case coalition of $k$ adversarial nodes as a graph cut problem. Our experimental findings are supported by a theoretical justification based on the scale-free topology of the LN.
SmartControllerJS: A JavaScript library to turn smartphones into controllers for web-based interactive experiments
Authors: Emma Poliakova, Fraser Dempster, Abubakr Mahmood, Jonathan Grizou
Abstract
We introduce SmartControllerJS, a new JavaScript library for fast, cost-effective designing of web applications controlled via everyday smartphones. At its core, SmartControllerJS establishes a connection between two webpages, one page running on a desktop browser and the other on the user's smartphone. The smartphone webpage loads a controller interface allowing users to control a web application running on their computer's browser. The SmartControllerJS framework enables fast iteration loops when designing interactive user experiments because it has minimal friction and allows for scaling, while having no running costs. We first describe how this library is built, how it can be used, and provide interactive examples. We then present two games designed for public screens along with results from user studies evaluating acceptability and ease of use. Finally, we implement a custom controller based on user feedback and introduce connection monitoring tools. We believe SmartControllerJS can accelerate the design of interactive experiments for researchers in Human-Computer Interaction, and be a useful tool for educational projects. To experience the various demos, we recommend reading this work on a desktop computer with your smartphone in hand. The library and the demos are available at https://github.com/SmartControllerJS
Keyword: out of distribution detection
There is no result
Keyword: out-of-distribution detection
There is no result
Keyword: expected calibration error
There is no result
Keyword: overconfident
There is no result
Keyword: overconfidence
There is no result
Keyword: confidence
There is no result
Keyword: scaling
Optimizing Age of Information with Correlated Sources
Mass Exit Attacks on the Lightning Network
SmartControllerJS: A JavaScript library to turn smartphones into controllers for web-based interactive experiments
Keyword: calibration
There is no result