Graduate Research Conference 2005 - Agenda

Department of Computer Science
Stony Brook University
Stony Brook, NY 11794-4400
10:00am - 5:30pm, Friday, April 22nd, 2005

1. Theory - 1306 Seminar Room: 10:00am - 12:00pm

- Theory of Computation
- Algorithms
- Logic, Artificial Intelligence, Machine Learning
- Special Topics: Natural Language Processing, Computational Biology, Computational Geometry

Organizer: Paul Fodor

Session Chair: Dhyanesh Damania

Judge: Prof. Anita Wasilewska

Theory Presentation 1 - time: 10:00am - 10:20am

- Title: On the complexity of finding circumscribed rectangles and squares of a Jordan curve
- Authors: Ker-I Ko, Fuxiang Yu (Presenter)
- Abstract: We consider in the real computation model of Ko and Friedman the problem of finding a rectangle or square that circumscribes about a Jordan curve. This problem is related to fixed point theory and root theory. We prove that for a polynomial-time computable Jordan curve, the complexity of finding a circumscribed square can be arbitrarily high. We also show that the complexity of finding the minimal area or perimeter of the circumscribed rectangles of a polynomial-time computable Jordan curve is characterized by the complexity class $\Sigma_2^P$. These results and some other related results build a link between geometry and the NP theory.
- Keywords: Complexity, Jordan curve, Circumscribed square, NP, Sigma_2^P.
- Link to the student webpage or the full content of the presentation:

Theory Presentation 2 - time: 10:20am - 10:40am (* Best Paper Award * )

- Title: Bounded Drift Iterated Snap Rounding
- Authors: Eli Packer
- Abstract: Snap Rounding and its varient, Iterated Snap Rounding, are methods for converting arbitrary-precision arrangements of segments into a fixed-precision representation. Each induces an output which might be unacceptable for cerain cases. While in Snap Rounding the distance between a vertex and a non-incident edge can be extremely small, inducing potential degeneracies, in Iterated Snap Rounding segments may drift far from their origin. We propose a new method based on the above two, Bounded Drift Iterated Snap Rounding, in which none of the above undesirable properties hold. As a result, we are the first to propose an algorithm whose output satisfies all the desirable properties discussed in the literature so far.
- Keywords: "computational geometry", "finite precision approximation", "robust computation"
- Link to the student webpage or the full content of the presentation: http://www.cs.sunysb.edu/~epacker/

Theory Presentation 3 - time: 10:40am - 11:00am

- Title: Meta-analysis based on control of False Discovery Rate
- Authors: Saumyadipta Pyne, Steve Skiena, Bruce Futcher
- Abstract: High-throughput technology such as microarrays can be used to simultaneously examine thousands of features, such as all the genes or regulatory sequences of an organism, and measure their expression under various experimental conditions. Two common requirements of microarray bioinformatics at this point are of techniques to combine the significance values for each feature across independent analogous experiments with high power, as well as to control the proportion of false positives among those features declared significant. In spite of the variety of existing methods to address either issue, in general the two issues, with dual requirements, are tackled independently and sequentially. We present a novel algorithm to simultaneously address and integrate the two, and discuss results due to its application.
- Keywords: Meta-analysis, False Discovery Rate, P-values, Microarrays
- Link to the student webpage or the full content of the presentation:

Theory Presentation 4 - time: 11:00am - 11:20am

- Title: Fast generation of DNA and protein motifs from profile matrices
- Authors: Paul Fodor, Joy Dutta, Steven Skiena
- Abstract: A DNA profile matrix gives probability of each base appearing in each position in a DNA sequence, so a length n profile matrix scores the likelihood of each possible DNA sequence of length n. The protein profiles are similar, the only difference is that the alphabet size varies with the number of different amino acids. Finding the most probable sequence is trivial, but finding the second most or k-th most probable sequence requires sorting 4n DNA sequences or 20n protein sequences. This paper describes and implements a geometric algorithm based on the zone theorem to provide a much faster alter-native to the brute force method.
- Keywords: "computational biology", "DNA", "protein", "motif", and "profile"
- Link to the student webpage or the full content of the presentation: http://www.algorithm.cs.sunysb.edu/compbio/motif/

 

2. Software - 2313A Conference Room: 10:00am - 12:30pm

- Programming Languages
- Compiler Design
- Database Systems and Web Services
- Graphics
- Special Topics: Verification

Organizer: Mohit Gupta

Session Chair: Amitabh Basu

Judge: Prof. CR Ramakrishnan

Software Presentation 1 - time: 10:00am - 10:20am  (* Best Paper Award * )

- Title: Efficient Modeling of Excitable Cells Using Hybrid Automata
- Authors: Pei Ye, Dept. of Computer Science, Stony Brook Univ. , Emilia Entcheva, Dept. of Biomedical Engineering, Stony Brook Univ., Radu Grosu, Dept. of Computer Science, Stony Brook Univ. , Scott A. Smolka, Dept. of Computer Science, Stony Brook Univ.
- Abstract: We present an approach to modeling complex biological systems that is based on Hybrid automata (HA). HA combine discrete transition graphs with continuous dynamics. Our goal is to efficiently capture the behavior of excitable cells previously modeled by systems of nonlinear differential equations. In particular, we derive HA models from the Hodgkin-Huxley model of the giant squid axon, the Luo-Rudy dynamic model of a guinea pig ventricular cell, and a model of a neonatal rat ventricular myocyte. Our much simpler HA models are able to successfully capture the action-potential morphology of the different cells, as well as reproduce typical excitable cell characteristics, such as refractoriness (period of non-responsiveness to external stimulation) and restitution (adaptation to pacing rates). To model electrical wave propagation in a cell network, the single-cell HA models are linked to a classical 2D spatial model. The resulting simulation framework exhibits significantly improved computational efficiency in modeling complex wave patterns, such as the spiral waves underlying pathological conditions in the heart.
- Keywords: Hybrid Automata, Action Potential, restitution.
- Link to the student webpage or the full content of the presentation: http://www.cs.sunysb.edu/~pye/CMSB05.pdf

Software Presentation 2 - time: 10:20am - 10:40am

- Title: Static Analysis of Atomicity for Programs with Lock-Free Synchronization
- Authors: Liqiang Wang, Scott D. Stoller
- Abstract: In concurrent programming, lock-free synchronization is very efficient but difficult to design correctly. This paper presents a static analysis to show that code blocks are atomic, i.e., that every execution of the program is equivalent to one in which those code blocks execute without interruption by other threads. Our analysis determines commutativity of operations based primarily on how synchronization primitives (including locks, load-linked, store-conditional, and compare-and-swap) are used. A reduction theorem states that certain patterns of commutativity imply atomicity. Atomicity is itself an important correctness requirement for many concurrent programs. Furthermore, an atomic code block can be treated as a single transition during subsequent analysis of the program; this can greatly improve the efficiency of the subsequent analysis. We demonstrate the effectiveness of our approach on several concurrent lock-free programs.
- Keywords: Atomicity, Lock-free, Non-blocking, Synchronization, Static Analysis, Verification, Linearizability.
- Link to the student webpage or the full content of the presentation: http://www.cs.sunysb.edu/~liqiang/lockfree.html

Software Presentation 3 - time: 10:40am - 11:00am

- Title: Conformal Colon Flattening
- Authors: Wei Hong, Xianfeng Gu, Feng Qiu, Arie Kaufman
- Abstract: We present an efficient colon flattening algorithm using conformal mapping, which is angle-preserving and the global distortion is minimized. The colon wall is segmented and extracted from the CT dataset first. The minuet handles are located and removed automatically. Moreover, our algorithm can handle high genus surfaces. Each vertex on the 3D colon surface is registered with a camera on the central path. The flattened colon image is then rendered using a direct volume rendering method accelerated with the GPU. Our algorithm is tested with a number of CT datasets of real pathological cases. We demonstrated the shape of the polyps is well preserved on the flattened colon images, which provides an efficient way to enhance the fly-through virtual colonoscopy system.
- Keywords: Conformal Mapping, Direct Volume Rendering, Virtual Colonoscopy
- Link to the student webpage or the full content of the presentation:

Software Presentation 4 - time: 11:00am - 11:20am

- Title: Answering Rule-Based Queries Efficiently with Complexity Guarantees
- Authors: Katia Hristova, Annie Liu
- Abstract: Many practical problems that may be difficult to understand and implement otherwise can conveniently and intuitively be expressed using rules. However, answering queries based on the rules efficiently with complexity guarantees has been a major challenge. In particular, existing implementations can have dramatically different running times depending on the order of rules or the order of hypotheses in a rule. We propose a method that combines magic set transformation and formal complexity analysis to automatically provide time complexity guarantees for answering queries efficiently, regardless of the orders of rules or hypotheses, and we apply the method to RBAC (role-based access control) and a Trust Management system for electronic health records.
- Keywords: Datalog, complexity analysis, security policy
- Link to the student webpage or the full content of the presentation:

Software Presentation 5 - time: 11:20am - 11:40am

- Title: Automated Type-Based Analysis of Data Races and Atomicity
- Authors: Amit Sasturkar
- Abstract: Concurrent programs are notorious for containing errors that are difficult to reproduce and diagnose at run-time. This motivated the development of type systems that statically ensure the absence of some common kinds of concurrent programming errors. This paper presents Extended Parameterized Atomic Java (EPAJ), a type system for specifying and verifying atomicity of methods in multi-threaded Java programs. Atomicity is a common correctness requirement for concurrent programs which requires that concurrent invocations of a set of methods be equivalent to performing the invocations serially in some order. EPAJ combines Flanagan and Qadeer's atomicity types with a new and significantly more expressive type system for analyzing data races, called Extended Parameterized Race-Free Java (EPRFJ), allowing a more accurate analysis of atomicity. The paper also presents a type discovery algorithm to automatically obtain EPRFJ types, and a static interprocedural type inference algorithm that, given EPRFJ types, infers atomicity types. These algorithms can be incorporated into testing and debugging tools, benefiting users who know nothing about type systems. We report our experience with a prototype implementation.
- Keywords: Data Races, Atomicity, Type Systems, Type inference
- Link to the student webpage or the full content of the presentation: http://www.cs.sunysb.edu/~amits/papers/atomicity-inference/

Software Presentation 6 - time: 11:40am - 12:00am

- Title: Incremental Points-To Analysis using logic programming
- Authors: Diptikalyan Saha, C. R. Ramakrishnan
- Abstract: Several program analysis problems can be cast elegantly as a logic program. In this paper we show how recently-developed techniques for incremental evaluation of logic programs can be refined and used for deriving practical implementations of incremental program analyzers. Incremental program analyzers compute the changes to the analysis information due to small changes in the input program rather than re-analyzing the program. We show the effectiveness of this approach by building a practical incremental context insensitive points-to analysis and evaluating this implementation for analyzing C programs with 10-70K lines of code. Experiments show that our technique can compute the changes to analysis information due to small changes in the input program in, on the average, 6\% of the time it takes to reanalyze the program from scratch, and with little space overhead.
- Keywords:
- Link to the student webpage or the full content of the presentation: http://www.lmc.cs.sunysb.edu/~dsaha/ppdp05.ps

 

3. Systems - 2311 Wireless Seminar Room:

Session 1: (10:00am - 12:30pm)

Topic: Wireless Networks

Organizer:  Ritesh Maheshwari

Session Chair:  Srikant Sharma

Judge: Prof. Tzi-cker Chiueh                 

 

Presentation 1 - Time: 10:00am - 10:20am

- Title: Beneficial Caching in Wireless Ad Hoc Networks
- Authors: Bin Tang, Samir Das, Himanshu Gutpa
- Abstract: In this paper we propose and design a benefit-based caching paradigm for wireless ad hoc networks. We start by designing a centralized algorithm in static network with provable bound under memory constraint of each node. We then implement it in a distributed manner and show it has a very close performance to the centralized one. Finally, we compare our algorithm with the latest published work in mobile environment. We first show that our proposed caching paradigm is more general and it can be easily extended into the peer-to-peer domain. Then, through extensive simulations in $ns2$, we show our scheme performs better in some range of parameters.
- Keywords: caching, ad hoc networks, algorithms
- Link to the student webpage or the full content of the presentation: http://www.cs.sunysb.edu/~bintang/papers/cache.pdf
 

Presentation 2 - Time: 10:20am - 10:40am

- Title: Solving Deafness and Directional Hidden Terminal Problem in Wireless Multihop networks using Directional Antennas
- Authors: Anand Prabhu Subramanian, Samir Das
- Abstract: The capacity of Wireless Multihop Networks is highly limited because the wireless channel is a shared one and a single transmission along a multihop path can inhibit a large number of nearby nodes. This is mainly due to the omnidirectional transmission that reserves a large region of space, while only a small portion of it is used for the communication. Other nodes defer their communication as they are exposed to this transmission. Directional Antenna is a technology that solves this problem and has a great potential to increase the capacity of multihop networks. Though directional antennas offer many benefits such as spatial reuse, coverage range, better link reliability and increase in network capacity, there are some tradeoffs that need to be addressed. Deafness and directional hidden terminal problem are two such problems that have a serious effect on network performance when left unaddressed. Deafness occurs when a transmitter fails to communicate to its intended receiver, because the receiver’s antenna is oriented in a different direction. The hidden terminal problem occurs when the transmitter failed to hear the RTS/CTS exchange between another pair of nodes and cause collision by initiating a transmission to the receiver node of the ongoing communication. Existing directional MAC protocols try to solve deafness but the directional hidden terminal problem is left unaddressed. In this paper, we present a novel directional MAC protocol that uses a single transceiver and a single channel and effectively solves both deafness and directional hidden terminal problem. In this scheme, we transmit the RTS and CTS packets omnidirectionally with the directional information about the following data transmission which is done directionally. Instead of transmitting RTS/CTS and data packets alternatively, we transmit a series of RTS/CTS packets in a control window before multiple simultaneous data transmissions take place in different directions. The length of the control window can be adaptively adjusted based on the traffic information to allow multiple data transmission simultaneously in different directions. Keywords: Directional Antenna, Control Window, Deafness, Hidden Terminal Problem
 

Presentation 3 - Time: 10:40am - 11:00am    (* Best Paper Award * )

- Title: Directional Antennas in Ad Hoc Networks
- Authors: Umesh Kumar
- Abstract: Most of the previous research on directional antennas in ad hoc networks assumes nodes have sufficient directional antennas (or an omni directional antenna) to cover all directions simultaneously. In this paper we consider networks where nodes maynot have enough directional antennas to cover all directions simultaneously. This may be caused due to antenna failures or deployment decisions (e.g monetary reasons). We slightly modify the approximation algorithms proposed in the literature for minimum degree spanning tree problem to determine whether the network can be connected with the available directional antennas at each node. We observe that even with the 3 antennas network is almost always connected ( more than 99.99\% times) and then evaluate how the performance of the network depends on the number and beamwidth of antenna's at each node using Qualnet simulator.
- Link to the student webpage or the full content of the presentation: http://www.wings.cs.sunysb.edu/~umesh

 

Presentation 4 - Time: 11:00am - 11:20am

- Title: Fault Tolerant Connected Sensor Cover with Variable Sensing and Transmission Ranges
- Authors: Zongheng Zhou, Samir R. Das, Himanshu Gupta
- Abstract: Sensor networks are often deployed in a redundant fashion. In order to prolong the network lifetime, it is desired to choose only a subsetof sensors to keep active and put the rest to sleep. In order to provide fault tolerance, this small subset of active sensors should also provide some degree of redundancy. In this paper, we consider the problem of choosing a minimum subset of sensors such that they maintain a required degree of coverage and also form a connected network with a required degree of fault tolerance. In addition, we consider a more general, variable radii sensor model, wherein every sensor can adjust both its sensing and transmission ranges to minimize overall energy consumption in the network. We call this the variable radii $k_1$-Connected, $k_2$-Cover problem. To address this problem, we propose a distributed and localized Voronoi-based algorithm. The approach extends the relative neighborhood graph (RNG) structure to preserve $k$-connectivity in a graph, and design a distributed technique to inactivate desirable nodes while preserving $k$-connectivity of the remaining active nodes. We show through extensive simulations that our proposed techniques result in overall energy savings in random sensor networks over a wide range of experimental parameters.
- Keywords: Sensor Network; Sensor Coverage; Topology Control;
- Link to the student webpage or the full content of the presentation: http://www.wings.cs.sunysb.edu/~zzhou/grc_2005.pdf

 

Presentation 5 - Time: 11:20am - 11:40am

- Title:  Fault-Tolerant Manycast to Mobile Destinations in Sensor Network
- Authors: Xianjin Zhu, Himanshu Gupta
- Abstract: Manycast is a group communication primitive wherein the source is required to send a data packet to a certain number/percentage of a given set of destinations. In this article, we design fault-tolerant protocols for the manycast operation in sensor networks when the given set of destinations are mobile. To develop efficient protocols, we propose a distributed location management scheme, which manages information about the location of mobile destinations in a distributed manner. Using the proposed location management scheme, we develop various fault-tolerant manycast routing protocols to reach the required number of destinations. First, we propose the rectangular-based manycast protocols using two routing schemes viz. rectangle flooding and GeoGrid-based routing. In addition, we develop a GridTree technique which is based upon developing fault-tolerant paths between adjacent nodes in an appropriately constructed geometric tree. We evaluate the performance of our designed techniques through extensive simulations.

 

Session 2: (2:00pm - 3:20pm)

Topics: P2P Networks, Security, File Systems

Organizer:  Abhishek Rai

Session Chair:  Ma Chi

Judge:  Prof Himanshu Gupta

Presentation 1 - Time: 2:00pm - 2:20pm

- Title: One-way Isolation: An Effective Approach for Realizing Safe Execution Environments
- Authors: Weiqing Sun, Zhenkai Liang, R. Sekar and V.N. Venkatakrishnan
- Abstract:In this paper, we present an approach for realizing a safe execution environment (SEE) that enables users to "try out" new software (or configuration changes to existing software) without the fear of damaging the system in any manner. A key property of our SEE is that it faithfully reproduces the behavior of applications, as if they were running natively on the underlying host operating system. This is accomplished via one-way isolation: processes running within the SEE are given read-access to the environment provided by the host OS, but their write operations are prevented from escaping outside the SEE. As a result, SEE processes cannot impact the behavior of host OS processes, or the integrity of data on the host OS. Our SEE supports a wide range of tasks, including: study of malicious code, controlled execution of untrusted software, experimentation with software configuration changes, testing of software patches, and so on. It provides a convenient way for users to inspect system changes made within the SEE. If the user does not accept these changes, they can be rolled back at the click of a button. Otherwise, the changes can be "committed" so as to become visible outside the SEE. We provide consistency criteria that ensure semantic consistency of the committed results. We also develop an efficient technique for implementing the commit operation. Our implementation results show that most software, including fairly complex server and client applications, can run successfully within the SEE. The approach introduces low performance overheads, typically below 10%.
- Keywords: Security, one-way Isolation, SEE.
- Link to the student webpage or the full content of the presentation: Not available
 

Presentation 2 - Time: 2:20pm - 2:40pm             (* Best Paper Award * )

- Title: An Efficient and Backwards-Compatible Transformation to Ensure Memory Safety of C Programs
- Authors: Wei Xu, Daniel C. DuVarney, and R. Sekar
- Abstract:Memory-related errors, such as buffer overflows and dangling pointers, remain one of the principal reasons for failures of C programs. As a result, a number of recent research efforts have focused on the problem of dynamic detection of memory errors in C programs. However, existing approaches suffer from one or more of the following problems: inability to detect all memory errors (e.g., Purify), requiring non-trivial modifications to existing C programs (e.g., Cyclone), changing the memory management model of C to use garbage collection (e.g., CCured), and excessive performance overheads. In this paper, we present a new approach that addresses these problems. Our approach operates via source code transformation and combines efficient data-structures with simple, localized optimizations to obtain good performance.
- Keywords:Memory Safety, C, Program Transformation.
- Link to the student webpage or the full content of the presentation: http://seclab.cs.sunysb.edu/seclab/pubs/papers/fse04.pdf
 

A Pairwise Currency Mechanism for Streaming in Non-Cooperative Environments

Presentation 3 - Time: 2:40pm - 3:00pm

- Title: A Pairwise Currency Mechanism for Streaming in Non-Cooperative Environments
- Authors: Vinay Pai, Kapil Kumar, Alexander E. Mohr
- Abstract:We investigate whether the application of a pairwise currency mechanism to mesh-based multicast overlay networks can give incentives to peers to contribute to the system, but at the same time maintain short propagation delays. Whereas traditional tree-based multicast systems are not amenable to pairwise currency mechanisms, we find that pairwise currency can be used to good effect in these newly developed unstructured mesh-based overlay multicast systems. This combination results in a simple, lightweight system in which only local state must be maintained. Through simulations and real-world experiments on PlanetLab, we show that nodes with upload to download ratios less than 1.0 are penalized by their neighbors and suffer significant packet loss, but nodes with ratios greater than 1.0 have minimal packet loss. Moreover, we show that when there are altruistic nodes willing to contribute their excess bandwidth to the system, it is feasible to accommodate nodes with limited upload capacity, such as ADSL users, and allow them to download without packet loss as well.
- Keywords:p2p, overlay multicast, incentives
- Link to the student webpage or the full content of the presentation: Not available
 

Presentation 4 - Time: 3:00pm - 3:20pm

- Title: Self-Optimizing OS Debugging: The Monte Carlo Approach
- Authors: Sean Callanan (presenter), Radu Grosu, Abhishek Rai (presenter), Scott Smolka, Mike True, Erez Zadok,
- Abstract:Large software is complex to debug, especially when it runs for long periods of time under production load, like an operating system kernel. Developers can include debugging with such software; however, debugging instrumentation often incurs overheads. Therefore, developers run their software in debugging mode for limited periods of time, and then release the software without debugging support. This leaves software vulnerable to those bugs that only appear after prolonged real-world use. We have developed a dynamic-analysis system that provides the best of both worlds: always include debugging instrumentation in commercially- shipped software, while reducing overheads by selectively turning off instrumentations at run-time.
Our system includes a kernel-specific compiler (KGCC) which instruments arbitrary parts of the kernel with checks against invariants. A novel Monte-Carlo model-checking algorithm which we developed estimates the confidence that a certain invariant holds. When the confidence exceeds a threshold, the inline monitoring instrumentation is turned off. Confidence for frequently-executed portions of the code will be accrued the fastest, leading to the rapid deactivation of instrumentation for those sections, which in turn restores performance to non-instrumented levels quickly.
- Keywords:Linux, debugging, instrumentation, model checking
- Link to the student webpage or the full content of the presentation: http://www.fsl.cs.sunysb.edu/~sean, http://www.fsl.cs.sunysb.edu/~abba