Optimizing Information Freshness in Wireless Networks
12:00 PM to 1:00 PM
CSB 480
Prof. Eytan Modiano, Laboratory for Information and Decision Systems, MIT
Age of Information (AoI) is a recently proposed performance metric that captures the freshness of the information from the perspective of the application. AoI measures the time that elapsed from the moment that the most recently received packet was generated to the present time. In this talk, we explore the AoI optimization problem in wireless networks.
We start by considering a wireless network with a number of nodes transmitting information to a Base Station. We develop three low-complexity transmission scheduling policies that attempt to minimize AoI, and evaluate their performance against the optimal policy. In particular, we develop a randomized policy, a Max-Weight policy and a Whittle’s Index policy, and show that they are guaranteed to be within a factor of two, four and eight, respectively, away from the minimum AoI possible.
We then extend our results to wireless networks under general interference constraints. We show that when fresh information is always available for transmission, a stationary scheduling policy is peak age optimal, and is within a factor of two of the optimal average age. When fresh information is not always available, and packet generation rate has to be controlled along with scheduling links for transmission, we prove an important separation principle: the optimal scheduling policy can be designed assuming fresh information, and independently, the packet generation rate control can be done by ignoring interference. Finally, we consider multi-hop wireless networks with general interference constraints, and obtain the optimal policy from the class of stationary policies in which links are activated according to a stationary probability distribution.
Add to GMail
Computer Security for Emerging Technologies
11:40 AM to 12:40 PM
CS Department 451
Earlance Fernandes, University of Washington
As our world becomes more computerized, security and privacy takes on a prominent role in allowing us to enjoy the benefits of new technologies without the risks. Addressing the new challenges that come with this role requires a change in how we approach and solve problems in computer security. My vision is that we must view computer security as a whole-system property ranging from the physical-layer right up to applications and end-users. In line with this vision, my approach to computer security involves formulating the right security problem to work on, addressing design-level issues by constructing strong defenses at the appropriate layer of abstraction, and challenging common assumptions to understand realistic threats. In this talk, I will give several examples of my approach and vision, focusing on emerging technologies that span the digital-to-physical interface. I will cover technical results at various level of abstraction, including analysis techniques that found exploitable design-level vulnerabilities in closed-source smart home platforms, a new design for trigger-action platforms that provides strong integrity guarantees, and an analysis of how realistic attacks on machine learning can occur in the physical world. Finally, I will share my vision of the future of security and privacy research in an increasingly connected world.
Hardware accelerators for deep learning: a proving ground for specialized computing
11:40 AM to 12:40 PM
CS Department 451
Brandon Reagan
The computing industry has a power problem: the days of ideal power-process scaling are over, and chips now have more devices than can be fully powered simultaneously, limiting performance. New architecture-level solutions are needed to continue scaling performance, and specialized hardware accelerators are one such solution. While accelerators promise to provide orders of magnitude more performance per watt, several challenges have limited their wide-scale adoption.
Deep learning has emerged as a sort of proving ground for hardware acceleration. With extremely regular compute patterns and wide-spread use, if accelerators can’t work here, then there’s little hope elsewhere. For accelerators to be a viable solution they must enable computation that cannot be done today and demonstrate mechanisms for performance scaling, such that they are not a one-off solution. This talk will present deep learning algorithm-hardware co-designs to answer these questions and identify the efficiency gap between standard hardware design practices and full-stack co-design to enable deep learning to be used with little restriction. To push the efficiency limits, this talk will introduce principled unsafe optimizations. A principled unsafe optimization changes how a program executes without impacting accuracy. By breaking the contract between the algorithm, architecture, and circuits, efficiency can be greatly improved. To conclude, future research directions centering around hardware specialization will be presented: accelerator-centric architectures and privacy-preserving cloud computing.
Add to GMail