All tutorials will be held on Friday, June 9, 2017, in the
National Center for Supercomputing Applications Building.
They are scheduled for three hours each, including a 20 minute break. Lunches are
Stein's Method for Steady-State Approximations: Error Bounds and Engineering Solutions
Friday, 9 am-noon, Room 1030 NCSA Building
Jim Dai (Cornell University)
Abstract: Through queueing systems modeling customer call centers, hospital
patient flows, and super market model, I will give an introduction on how to use Stein's method both as an engineering tool for generating good steady-state approximations and as a mathematical too for establishing error bounds for these approximations. These approximations are often universally accurate in multiple parameter regions, from underloaded to overloaded (when abandonment is possible). I will focus on diffusion models for performance analysis, briefly discussing recent works by others on ergodic optimal controls and mean-field approximations. The talk is based on joint works with Anton Braverman and Jiekun Feng from Cornell and Pengyi Shi from Purdue.
Security Economics: From Game Theory to Field Measurements
Friday, 9 am-noon, Room 2100 NCSA Building
Nicolas Christin (Carnegie Mellon University)
Abstract: In a world where a majority of online security attacks are carried out
for financial profit, understanding incentives of both attackers and
targets has become critical to strengthening security. This tutorial
will present a broad overview of recent work in security economics
attempting to untangle these incentives. In the first third of the tutorial,
I will introduce some of the more theoretical work that was developed
in the early to mid-2000s. Most of this pioneering research was based
on game-theoretical models capturing negative externalities among
defenders and their impact on network security. However, it became
quickly apparent that behavioral biases also greatly affect defensive
security posture. I will thus discuss and illustrate some of these
biases in the second third of this talk, and explain the relationship
between computer security, privacy and behavioral economics. Finally,
the last third of the tutorial will deal with attacker incentives.
Through case studies of unlicensed prescription drug sales and online
anonymous markets, I will present measurements and data analyses of
miscreant behavior that yield insights on possible, economically
effective, defenses. I will conclude by outlining a roadmap for security
research combining measurements, mathematical modeling and behavioral
Routing, Scheduling, and Networking in Data Centers
Friday, 1:30-4:30 pm, Room 1030 NCSA Building
R. Srikant and
(University of Illinois)
Abstract: We will discuss many examples of resource allocation problems that arise in data centers and present an overview of the analytical tools used to study them. The common theme is one of load balancing tasks among multiple servers in a data center. In one variant of the load balancing problem, each task has to be assigned to one server among a very large number of servers with the goal of minimizing task completion times, while ensuring that the complexity of sampling the server loads is small. In another variant, tasks arrive in batches called jobs, and the goal of load balancing is to ensure small job completion times rather than small task completion times. We will also consider data locality issues that arise due to the fact that a task may incur smaller processing delays at some servers while it may incur larger processing delays at other servers, depending upon whether the data required to execute a task is available locally at a server or not. Finally, we will discuss flow-level resource allocation problems that arise in data center networks due to the need for servers to communicate with each other.
Routing Money, Not Packets: A Tutorial on Internet Economics
Friday, 1:30 - 4:30 pm, Room 2100 NCSA Building
Vishal Misra (Columbia Univerity)
Richard Ma (National University of Singapore)
Abstract: This tutorial is in the broad area of Internet Economics, specifically applying ideas from game theory, both Cooperative and Non-Cooperative. We consider the origins of the Internet architecture, and the evolution of the Internet ecosystem from a protocol and application standpoint. We next look at the evolution of the pricing structure on the Internet along three different dimensions: (a) between
ISPs, (b) between ISPs and content providers, and (c) between ISPs and end users. We present mathematical models describing the pricing structures in each dimension, the interaction between the three and competition amongst the entities leading to the notion of Network Neutrality. We look at various definitons of Network Neutrality and analyze the the impact of mechanisms like paid peering, zero rating and differential pricing on the principle of Network Neutrality.