Contact Information

Murat Alanyali

"A Distributed Architecture for Decision Making in Sensor Networks"

Presentation (.pdf)

We consider decision making based on noisy measurements collected by networked sensors. The centralized solution of the problem, in which all measurements are available to the decision maker, is well-understood, however it is not suitable for envisioned applications of sensor networks due to scalability issues. Here we focus on decentralized solutions based on local message passing by the sensors. It is assumed that, in addition to taking measurements, each sensor can communicate with several other sensors.  We study variants of Pearl's belief propagation algorithm, and obtain conditions under which local message passing leads to a consensus in which each sensor eventually identifies the centralized solution of the problem.



Azer Bestavros

"Exploiting the Transients of Adaptation for RoQ Attacks on Internet Resources"


Over the past few years, Denial of Service (DoS) attacks have emerged as a serious vulnerability for almost every Internet service. An adversary bent on limiting access to a network resource could simply marshal enough client machines to bring down an Internet service by subjecting it to sustained levels of demand that far exceed its capacity, making that service incapable of adequately responding to legitimate requests.  In this talk I will expose a different, but potentially more malignant adversarial attack that exploits the transients of a system's adaptive behavior, as opposed to its limited steady-state capacity. In particular, I will show that a determined adversary could bleed an adaptive system's capacity or significantly reduce its service quality by subjecting it to an unsuspicious, low-intensity (but well orchestrated and timed) request stream that causes the system to become very inefficient, or unstable.  I will give examples of such "Reduction of Quality" (RoQ) attacks on a number of common adaptive components in modern computing and networking systems. RoQ attacks stand in sharp contrast to traditional brute-force, sustained high-rate DoS attacks, as well as recently proposed attacks that exploit specific protocol settings. I will present numerical and simulation results, which are validated with observations from real Internet experiments.

This work was done in collaboration with Mina Guirguis and Ibrahim Matta.



Ken Birman

"Navigating in the Dark: New Options for Building Self-Configuring Embedded Systems"

Presentation (.ppt)

Far too often, we build computing systems in ways that require delicate hand-configuration and parameter tuning. The resulting platforms are often expensive to deploy and fragile. With the growing use of wireless systems, small sensors, embedded computing nodes, and very large dynamically established collaborations (e.g. clients of a web services platform), we are seeing applications for which this traditional approach is just unworkable. Astrolabe, a tool developed by our group at Cornell, can greatly simplify this task by robustly capturing system configuration data, online performance metrics, or even application-specific state, track it as it evolves, and help the user data mine to make sense of what may be a very large amount of distributed state information. The approach scales well, is secure, and is astonishingly robust against disruption. We see Astrolabe as a mechanism for helping loosely coupled clients coordinate their behavior in large configurations where, traditionally, each node was on its own and forced to navigate in the dark. Astrolabe can run on Windows or Linux, and has a small footprint, and uses very limited computational and communication resources to accomplish its task.


Tony Ephremides

"ENERGY: The Root of All Pervasiveness"

Presentation (.ppt)

In the wireless domain, and especially for the case of ad hoc networks, the key to pervasive computing and networking is the uninterrupted supply of adequate amount of energy for processing and transmission.  In this talk we will focus on a number of key ideas related to energy efficiency and on some results that illustrate the trade-offs between energy conservation and performance that may be necessary.  The key ideas are that energy can be saved by not only improving the hardware but also by carefully designing network protocols at all layers. Related to that is the notion of vertical layer integration, the exploitation of which can assist in realizing energy savings. And, last but not least, the need to balance energy expenditures for processing against energy consumption for transmission. Especially in the context of sensor networks, where the objectives of long lifetime and high performance are directly conflicting, we will develop a more detailed model that illustrates the need for compromises and present some preliminary results that raise key design criteria.



Johannes Gehrke

"Energy-efficient Data Management in Sensor Networks"


In the Cornell Cougar Project, we investigate a database approach to sensor networks: Clients "program" the sensors through queries in a high-level declarative language (such as a variant of SQL). The first part of the talk describes a hybrid push-pull model for query processing, in which data is pushed to "view" nodes, from where the data is pulled by  queries.  We combine the push-pull model with novel multi-query optimization  methods for aggregate queries, resulting in a family of distributed algorithms that greatly reduce energy usage. The second part of the talk focuses on a novel integrated MAC and  routing layer protocol.  The protocol avoids all collisions at the MAC layer by turning radios on only during specified time intervals, and it contains a routing protocols that is tuned to the  MAC layer schedule.  Experiments show that we achieve significant energy savings at the cost of higher latency. This is joint work with Al Demers, Rajmohan Ramarajan, Niki Trigoni, and Yong Yao.



Mario Gerla

"Opportunistic networking and sensing in the urban vehicle grid"

Presentation (.ppt)

While the most popular examples of ad hoc networking (battlefield, disaster recovery, homeland defense, etc) are mutually exclusive with the notion of "infrastructure networks and the Internet", it is a fact that those standalone, self configured ad hoc networks still have a hard time finding commercial applications and touching people's everyday lives. An emerging concept today that probably will reverse this trend is the notion of "opportunistic ad hoc networking" - basically an ad hoc (non infrastructure) subnet that attaches to the Internet via a wireless infrastructure link like a Wireless LAN, a 2.5/3G link or a satellite connection. Examples range from kids playing Internet games in a shopping mall; to nomadic users who commit their packets to the air in the urban grid, so that they are picked up by a passing bus, or a car, and delivered to the nearest Mesh Network access point. In this talk, I will discuss the advantages and challenges posed by this "opportunistic ad hoc networking". I will focus on the urban "vehicular grid", seen as a dynamic, dense, large scale ad hoc network. I will discuss applications of urban grid "opportunistic ad hoc networking" to car navigation safety; entertainment, and; homeland defense. For the last application, I will outline the opportunities (and challenges) presented by the urban grid viewed as a sensor network.



Guerney  Hunt

"Integrating Pervasive Computing into the emerging Utility Infrastructures"


The objective of a utility infrastructure is to provide computing services to its customers. Within a corporation, the customers are its employees. Outside the corporate environment, aggressive ISPs and ASPs are moving towards a utility infrastructure as mechanism to provide services for their customers. Traditionally service orientated computing was thought of as a new model enhance the utilization of traditional computing resources like servers and business applications. However, to be effective utility computing has to reach all the way into the telecommunications network. Utility computing allows dynamic provisioning and assignment of resources to users. Therefore Pervasive Computing has to integrate into this model as an effective way to deliver services to customers. This paper will explore what this environment will look like and the challenges of integrating into a pay as you go infrastructure.



Bhaskar Krishnamachari

"Self-Optimization in Sensor Networks"

Presentation (.ppt)

Wireless sensor networks are expected to be significantly resource-limited in most scenarios, particularly in terms of energy. In recent years, researchers have advocated and studied cross-layer design techniques as the primary methodology to leverage application-specifity for optimizing system performance. We argue that another powerful design principle is to make sensor networks autonomously learn application-specific information through sensor and network observations during the course of their operation, and use these to self-optimize system performance over time. We discuss several example application scenarios where such self-optimization can be used, including target localization, model-based compression, and data delivery to mobile sinks.



Jim Kurose

"Networking Research Challenges in Collaborative Adaptive Sensing of the Atmosphere"


There is a wide range of sensor networks, ranging from small, power-constrained embedded sensing networks (used, e.g., for monitoring structural response or contaminant transport in the environment) to large, high-bandwidth, powered remote sensing networks (e.g., meteorological radar networks, and video sensing networks). Despite their obvious differences, these sensor networks also have many similarities and share a number of common technical challenges. In this talk, we consider the class of large, high-bandwidth, powered remote sensing networks, and identify the networking research challenges posed by such systems. We also consider the similarities and differences between these systems and their small, power-constrained embedded sensing counterparts. We illustrate these issues drawing on our experiences in designing and implementing a dense network of meteorological radars whose goals is to collaboratively and adaptively sense the lowest few kilometers of the earth's atmosphere.



Eytan Modiano

"Cooperative routing in wireless networks"


We attempt to extend Multiple input and multiple output (MIMO) techniques to network layer routing. We consider both transmission-side diversity and receiver-side diversity. We assume that each node in the network is equipped with a single omni-directional antenna and multiple nodes are allowed to coordinate their transmissions to achieve transmission-side diversity. The problem of finding the minimum energy route under this setting is formulated and asymptotic results are obtained for the resulting energy savings for both a regular line network topology and a grid network topology (of 39% and 56% respectively). Finally we present some preliminary results demonstrating that receiver-side diversity can be used to reduce outage probability when communicating over block fading channels.



Yoram Ofek

"The Trusted Flow Protocol for Remotely Authenticated and Trusted Operation"

Presentation (.pdf)

Assuring that a software module execution is correct in the sense that the user faithfully executes a given code with defined parameters and constraints is an open problem, which is especially important in the context of: (1) executing network protocols, (2) computing over networks, and (3) protecting (audio/video) content delivered over networks.  In this work, we present a novel software protection method to these problems by facilitating remotely authenticated operation. This is achieved by assuring that the software is not changed prior to and during execution. The solution is based on continuously emanating a flow of idiosyncratic signatures from the executed software modules. These signatures authenticate the software modules from which they have emanated. The idiosyncratic signatures are generated by a secret function that is hidden (e.g., obfuscated) in the software and whose execution is subordinated to the proper execution of the software modules being authenticated. The flow of signatures is validated at a remote trusted component. This generation and validation method of idiosyncratic signatures is called TrustedFlow protocol.  The TrustedFlow protocol is a general add-on protection method that complements other security tools, such as, trusted computing, IPSec, SSL. The TrustedFlow protocol has broad potential applications in: grid-computating, intrusion and denial of service avoidance, digital right management of audio and video content, and the protection of networks and servers from misbehaving client programs (e.g., TCP, SLA).  Some typical scenarios that can benefit from the TrustedFlow protocol will be discussed:  

(*) Trusted digital right management (DRM): assuring that client media application that receive (copyright protected) audio/video content content is handling it in accordance to agreed upun usage rules as specified, for example, by XrML.  

(*) Denial of service (DoS) avoidance: assuring that trusted clients receive premium service even when the network/server is under DoS or distributed DoS attack. This is especially critical for premium streaming media applications, such as, voice/video over IP, through firewalls, while minimizing delay and maximizing throughput.  

(*) Trusted distributed computing: assuring in peer-to-peer collaborative computing, such as, grid-computing, that the computation on one machine can be mutually authenticated by the other.  

(*) Trusted wireless infrastructure: protecting mobile devices that are roaming from one place to another from misbehaving servers and (ad-hoc) access points. Often the issue of authentication of mobile wireless devices is limited to the problem of a wired server and access point checking the identity of a wireless handheld device communicating with it. Nevertheless, the authentication issue also includes the dual problem of a mobile device checking the identity of a server or access point. Moreover, identity verification is not sufficient to ensure sufficient security levels since guarantees on the identity of a server or access point do not imply any guarantees on the software being executed by the (ad-hoc) access point during run-time. In fact, untrustworthy software could be executed on a properly authenticated access point. Being able to trust software being remotely executed is particularly important for mobile devices that have limited processing, storage, and power resources



Sumit Roy 

"Maximizing Aggregate Throughput in 802.11 Mesh Networks with Physical  Carrier Sensing and Two-Radio Multi-Channel Clustering"

Presentation (.ppt)

Increasing aggregate throughput in a mesh network (multi-hop network with fixed but randomly deployed nodes) is a primary concern for enabling the vision of a digital home with support for multimedia streams. Spatial reuse in such networks can allow multiple communications to proceed simultaneously, hence proportionally improve the overall network throughput.  To maximize spatial reuse, the MAC protocol must enable simultaneous co-channel transmitters at a separation distance that is sufficient to mitigate co-channel interference. Within that distance, a set of orthogonal channels can be employed by other users to further increase the aggregate throughput.


This paper reports on two contributions towards increasing the aggregate throughput in 802.11 mesh networks. It is shown that physical carrier sensing enhanced with a tunable sensing threshold is effective at mitigating co-channel interference in 802.11 mesh in scenarios where virtual carrier sensing (via RTS/CTS) is insufficient.


Moreover, for multi-channel .11 mesh networks, architecture for channel clustering based on two-radio nodes is proposed. Distributed clustering is achieved using the Highest-Connectivity Cluster (HCC) algorithm. All inter-cluster communications are performed on a common channel using the primary radio, while intra-cluster communications use the secondary radio with the channel selected by a new Minimum Interference Channel Selection (MIX) algorithm that minimizes the co-channel interference (CCI). Backward compatibility is guaranteed by allowing a legacy single-channel device connect to the new two-radio device through the primary radio. Simulation results for large-scale 802.11b and 802.11a networks demonstrate the significant improvement in one-hop aggregate throughput. Specifically, the new two-radio multi-channel mesh solution increases the performance of 802.11b mesh by more than twice throughput of the traditional single-radio single-channel mesh.


Andreas Savvides  

"Experience from the Design and Deployment of a Self-Configuring Beaconing System"


This talk describes our experience with the development of a rapidly installable self-configuring beaconing system for a smart environment. The beaconing system is comprised of a set of custom-designed wireless sensor nodes that act as smart beacons that help other wireless devices in the room to accurately localize with a few centimeters of accuracy. The novel feature of the smart beacons system is that it can self-calibrate in a few seconds after its deployment and readily provide reference information helping other sensors in the room to infer their locations. The provided self-configuration ability results in low cost rapidly installable system used to provide location information in a deeply instrumented classroom setup. In this talk we overview the main system components and localization algorithms and report on our experiences and lessons learned during the design and use of this system.



Fred Schneider

"Trustworthy Services from Untrustworthy Components:  Overview"

Presentation (.ppt)

It is indeed possible to build a trustworthy system from untrustworthy components, but not by implementing the fault-tolerance and security separately. Separation of concerns just does not apply. This lecture will discuss the role of threshold cryptography and weak computing models in building fault-tolerant and secure services. We will also discuss basic protocols.



Ness Shroff

"On the Fundamental Relationship Between Achievable Capacity and Delay in Mobile Wireless Networks"


In this work, we establish the fundamental relationship between the achievable capacity and delay in mobile wireless networks. Under an i.i.d. mobility model, we first obtain the following upper bound on the achievable capacity. For a mobile wireless network with n nodes, if the per-bit-averaged mean delay is bounded by D, then the upper bound on the per-node capacity is on the order of the  cubic-root of D/n (up to a logarithmic factor). By studying the condition under which the upper bound is tight, we are able to identify the optimal values of several key scheduling parameters. We then develop a scheduling scheme that can almost achieve  the upper bound up (up to a logarithmic factor). This indicates that the upper bound is tight. Our scheduling scheme also achieves a provably larger per-node capacity than schemes reported in previous works. In particular, when the delay is bounded by a constant, our scheduling scheme achieves a per-node capacity that is inversely  proportional to the cubic-root of n (up to a logarithmic factor). This  implies that for the i.i.d. mobility model, mobility improves that achievable capacity of static wireless networks, even with constant delays! Finally, the insight drawn from the upper bound allows us to identify limiting factors in existing scheduling schemes. These results present a relatively complete picture of the achievable capacity-delay tradeoffs under different settings.



Martha Steenstrup

"Emergent Structure among Self-Organizing Devices"

Presentation (.ppt)

The goal of pervasive computing and networking is to provide users with spontaneous opportunistic access to computational and communications resources independent of current location.  These resources belong to autonomous devices distributed over a wide area and operating collectively. The sheer number of devices and their heterogeneity in capabilities and capacities, however, implies that a single device is unlikely to be cognizant of the current state of all others. Thus, to meet the diverse needs of multiple simultaneous users, the algorithms for controlling access to and use of resources must be distributed over devices, enabling them to make local decisions based on partial information but which are likely to have global ramifications. We illustrate the utility of this approach with several examples of emergent structure among self-organizing mobile wireless devices, arising from solutions to the problems of frequency assignment, topology control, and route selection.



Sergio Servetto

"The System Architecture of a Distributed Transmitter for the Sensor Reachback Problem based on Radar Signals"

Presentation (.pdf)

We consider the problem of reachback communication in sensor networks. In this problem, a large number of sensors are deployed on a field, to measure the state of some physical process that unfolds over the field, who then need to cooperatively send this information back to a distant receiver for further processing.  We formulate the problem as a multiple-input, single-output (MISO) system, and develop a time-division scheme based on transmission of simulated radar echoes.  Information is encoded in the spatial electromagnetic reflectivity function of virtual point reflectors, and decoded with a conventional range radar receiver. Transmitter diversity and the use of pulse compression radar waveforms are exploited for both increased reliability and increased data rate. Information theoretic and simulation-based performance characterizations are also presented.


Pramod K. Varshney

"On Sensor Networking and Signal Processing for Smart and Safe Buildings"

Presentation (.ppt)  



Moti Yung

"Security and Privacy in Activeworld"


Activeworld is the coming setting where a user keeps mobile devices that emit signals most of the time. This situation provides for challenges from security and privacy point of view. Advantages and disadvantages as well as research challenges will be discussed.



Home | CPCN Home | Program | Abstracts | Speakers | Committees | Accommodations | Directions | Contact Information

Rensselaer's Center for Pervasive Computing and Networking (RCPCN), 110 8th St., Troy, NY 12180. (518) 276-2714
Copyright 2003
Rensselaer's Center for Pervasive Computing and Networking.  All rights reserved worldwide.

For problems or questions regarding this web contact szymansk@cs.rpi.edu
Last updated: 04/19/04.