Proceedings of the 1996 ICPP Workshop on Challenges for Parallel Processing, 1996
Many tree-based multicast algorithms have been proposed to provide an e cient software implementa... more Many tree-based multicast algorithms have been proposed to provide an e cient software implementation on parallel platforms without hardware multicast support. These algorithms are either architecture-dependent (not portable) or architectureindependent (portable) but do not provide good performance when porting to di erent parallel platforms. Based on the LogP model, the proposed parameterized communication model can more accurately characterize the communication network of parallel platforms. The model encompasses a number of critical system parameters which can be easily measured on a given parallel platform. Based on the model, e cient methods to construct optimal multicast trees are proposed for both 1-port and -port communication architectures. Experimental results conducted on the IBM/SP at Argonne National Laboratory are presented to compare the performance of the optimal multicast tree with two other known tree-based multicast algorithms. We claim that our proposed multicast algorithms can be ported to di erent parallel platforms and provide a near-optimal performance as the truly machine-speci c optimal performance is achievable only when the underlying detailed network characteristics are considered.
2006 International Conference on Collaborative Computing: Networking, Applications and Worksharing, 2006
Sensors are envisioned to be at the center of distributed collaborative computing services involv... more Sensors are envisioned to be at the center of distributed collaborative computing services involving time-critical decision support. Sensors are small devices with limited communication and computational capabilities that collect data on their neighboring physical world and send the data periodically to server machines. Sensors form a collaborative network with these servers, where the sensors gather information and the servers perform various operations (e.g. filter, aggregate, join etc) on the information streams in real-time according to predefined queries or rules. Sensor data streams are continuous, un-ending and have highly volatile characteristics. As a result, traditional database systems are inappropriate for handling queries for sensor streams, and several stream data management systems have been proposed in the literature. In this paper we focus on a special type of query, namely join queries, which is the most expensive query operator. Here, we address the problem of finding an optimal join tree that maximizes throughput for sliding window based multi-join queries over continuous sensor data streams. We present a polynomial time algorithm Fodp and three variants of Fodp. Our experiments in ARES 1 show that for almost all instances, trees from Fodp and its variants perform close to the optimal trees from our exponential time algorithm OptDP [15], and significantly better than existing XJoin based heuristic algorithms.
Proceedings.Twenty-First Annual Joint Conference of the IEEE Computer and Communications Societies, 2002
Network survivability is a crucial requirement in high-speed optical networks. Typical approaches... more Network survivability is a crucial requirement in high-speed optical networks. Typical approaches of providing survivability have considered the failure of a single component such as a link or a node. In this paper, we consider a failure model in which any two links in the network may fail in an arbitrary order. Three loopback methods of recovering from double-link failures are presented. The first two methods require the identification of the failed links, while the third one does not. However, precomputing the backup paths for the third method is more difficult than for the first two.
Proceedings IEEE INFOCOM 2006. 25TH IEEE International Conference on Computer Communications, 2006
As the commercial success of the IEEE 802.11 protocol has made wireless infrastructure widely dep... more As the commercial success of the IEEE 802.11 protocol has made wireless infrastructure widely deployed, user organizations are increasingly concerned about the new vulnerabilities to their networks. While various security issues have been extensively studied, the threats posed by denial-of-service (DoS) attacks have not been fully exploited. In this paper, we consider DoS attacks posed by cheating on the backoff rules in the IEEE 802.11 DCF protocol and propose a scheme detecting such adversaries. Our scheme is based on the sequential hypothesis testing. We first develop analytical models for packet interarrival time distribution from each station in the network where multiple cheating stations co-exist. Using the characterization of this probability distribution, we develop an algorithm to detect cheating stations based on the throughout degradations observed at normal stations. Our simulation results show that the proposed algorithm only requires very small number of observations of packets with very small value (i.e., less than 0.1%) of false positive and false negative decisions. That is, our proposed algorithm performs significantly fast and also accurately.
With the increasing popularity of the World Wide Web (Web), it is very difficult for a single pop... more With the increasing popularity of the World Wide Web (Web), it is very difficult for a single popular Web server to handle the explosive demand from its clients. Web caching (proxy) is an effective technique to reduce a Web server's load, alleviate Internet traffic, and reduce user's response time when accessing a Web service. A fundamental question is where to
This paper considers the problem of allocating files in a document tree among multiple processors... more This paper considers the problem of allocating files in a document tree among multiple processors in a parallel webserver. It is assumed that access patterns are characterized by branching probabilities for an access that starts at a node and progresses down the tree. A combinatorial optimization problem is formulated that includes load balancing and communication costs. The general problem is
Our proposed scheme seeks to defend flooding- based distributed denial-of-service (DDoS) attacks ... more Our proposed scheme seeks to defend flooding- based distributed denial-of-service (DDoS) attacks in the Internet. An easy, yet very disruptive, way to cause unfairness to the legitimate users is to deplete the network bandwidth by sending high rate unresponsive flows from multiple sources. The network congestion created by such malicious flows causes most legitimate packets dropped at routers without reaching
This paper considers the problem of maximizing throughput in a multi-carrier wireless network tha... more This paper considers the problem of maximizing throughput in a multi-carrier wireless network that employs predictive link adaptation. The contributions of this paper are the following. We explicitly consider the time-penalty induced by the transmission-mode changes in wireless networks. We study the trade-offs between the channel-state prediction quality and the throughput. We provide sound offline problem analysis, with optimal or approximate solutions to the extent possible. We consider online version of the wireless channel scheduling problem, and provide a lower bound and propose algorithms that guarantee competitive ratios. We also extend our algorithms to accommodate the model when the actual channel capacity is different from as predicted earlier. Our results show that a modest consumption of resources for channel prediction and link adaptation may result in a significant throughput improvement, with only marginal gains through further enhancement of the prediction quality.
First International Conference on Broadband Networks, 2004
Channel-aware scheduling and link adaptation (LA) methods are widely considered to be crucial for... more Channel-aware scheduling and link adaptation (LA) methods are widely considered to be crucial for realizing high data rates in wireless networks. Multi-carrier systems that spread information bits over the entire signal band can take advantage of the frequency selective fading, and choose the sub-carrier(s) that have the best channel conditions for transmission. However, predicting the future channel states, and adjusting the transmission schedules and parameters accordingly, may consume valuable system resources, such as bandwidth, time, and power. Sub-carrier grouping, which refers to monitoring/treating a set of adjacent sub-carriers as a single unit, has been proposed to reduce the overhead associated with LA. This paper considers various models for sub-carrier grouping, each representing a different level of system complexity, and investigates the trade-offs between the sub-carrier grouping granularity and the link throughput. We first present an offline dynamic programming algorithm for finding an optimal solution, assuming perfect estimates of the channel states. The solution obtained through this formulation can serve as a performance bound for online algorithms. We then propose an online algorithm for the case of perfect channel state estimates, which is shown to have very close performance to the off-line optimal solution. This online algorithm is also extended to incorporate imperfect channel state estimates due to (i) variations in channel states among the sub-carriers within the same group, and (ii) prediction inaccuracies in estimating the future time-slots. Performances of the algorithms are evaluated through simulations and comparisons with the off-line solution. The trade-off between the sub-carrier grouping granularity and the link throughput is also presented. Overall, our results can provide some guidelines from the network scheduling perspective for deciding what kind of sub-carrier grouping model and granularity to use in multi-carrier networks with LA.
Channel-aware scheduling and link adaptation methods are widely considered to be crucial for real... more Channel-aware scheduling and link adaptation methods are widely considered to be crucial for realizing high data rates in wireless networks. However, predicting the future channel states, and adjusting the transmission schedules and parameters accordingly, may consume valuable system resources, such as bandwidth, time, and power. This paper considers the trade-offs between the prediction quality and the throughput in a wireless network that uses link adaptation and channel-aware scheduling. In particular, we study the effects of the look-ahead window, i.e., the range of future time slots on which we have channel state estimates, and the reliability of the channel state estimates on the throughput. We develop an online scheduling algorithm for a multi-channel multi-user network that employs predictive link adaptation, and generalize it to incorporate imperfect channel state estimates. We use this heuristic together with performance bounds to the offline version of the problem to evaluate the performance with varying prediction quality. Our results suggest that it may be possible to reap most of the potential channel-aware scheduling benefits with a small look-ahead, and imperfect channel state estimates. Thus, a modest consumption of resources for channel prediction and link adaptation may result in a significant throughput improvement, with only marginal gains through further enhancement of the prediction quality. Our results can provide meaningful guidelines in deciding what level of system resource consumption is justified for channel quality estimation and link adaptation.
2007 IEEE International Conference on Communications, 2007
CDMA 1x Ev-DO is an evolution of the CDMA2000 3G wireless standard to enable high rate packet dat... more CDMA 1x Ev-DO is an evolution of the CDMA2000 3G wireless standard to enable high rate packet data services up to 2.4Mbps in Rev 0, 3.1Mbps in Rev A, and 4.9Mbps in Rev B. To support such high data rates, CDMA 1x Ev-DO utilizes TDMA technology on the downlink allowing a single receiver per time slot. Accordingly, a scheduling algorithm is necessary to determine which user receives data in a given time slot. Typically, schedulers consider efficiency (increasing network throughput) or fairness (fair allocation of resources among users) as the basis for scheduling. To realize a reasonable tradeoff between efficiency and fairness, opportunistic schedulers have been proposed that take advantage of instantaneous improvements in users radio conditions by serving them at higher rates. However, given the increasing demand to support user QoS requirements, a QoS-aware scheduler may need to serve users even at times of inadequate radio conditions. In this paper, we propose a QoS-aware packet scheduling algorithm that takes into account policy rules that govern the relationships between different user QoS classes. Our scheduler uses marginal utility functions defined to embody the given rules when selecting users to be served on the downlink. Our scheduling algorithm is implemented in the OPNET module of the CDMA 1x Ev-DO system that we have developed, and its performance is compared to other scheduling algorithms. The simulation results show that our scheduler performs well ensuring the policy rules are followed, meeting users' QoS requirements, and providing fairness among users within the same QoS class.
2007 Fourth International Conference on Broadband Communications, Networks and Systems (BROADNETS '07), 2007
This paper considers the problem of maximizing the system throughput in IEEE 802.16 broadband acc... more This paper considers the problem of maximizing the system throughput in IEEE 802.16 broadband access networks with mesh topology, and the following results are presented. We first consider a simplified linear network with only uplink traffic and provide an optimal scheduling algorithm and establish an analytical result on the length of the schedule. We then consider the problem of routing and packet scheduling in general topology, and show its NPcompleteness. We also provide an ILP formulation for this problem. Based on our optimal algorithm for linear networks, we propose algorithms that find routes and schedules of packet transmissions in general mesh topologies. The performance of our proposed algorithms is analyzed using the NS-2 simulator. The results show that the suggested algorithms perform significantly better than other existing algorithms.
This paper studies a model for matching and scheduling a task structure onto a heterogeneous suit... more This paper studies a model for matching and scheduling a task structure onto a heterogeneous suite of computers. The paper first generalizes Optimal selection theory (OST) and its later augmentations to include general dependency graphs, which represent dependency between subtasks, and presents the Generalized optimal selection theory (GOST) model. The GOST model allows non-optimal choices of machines, as in AOST, heterogeneous code blocks, as in HOST, and incorporates communication time, system reconfiguration time, and data reformatting time (needed when data is exchanged between heterogeneous processors). Under the GOST model, we address the problem of matching/scheduling dependency graphs on a heterogeneous suite of computers. For arbitrary dependency graphs, matching/scheduling is NP-complete. But for the special instances of trees, and, more import.ant,ly, in the case of series-parallel dependency graphs we present polynomial time algorithms for optimal matching/scheduling . a 0-8186-5592-5/94 $3.00 0 1994 IEEE
Proceedings of the 1996 ICPP Workshop on Challenges for Parallel Processing, 1996
Many tree-based multicast algorithms have been proposed to provide an e cient software implementa... more Many tree-based multicast algorithms have been proposed to provide an e cient software implementation on parallel platforms without hardware multicast support. These algorithms are either architecture-dependent (not portable) or architectureindependent (portable) but do not provide good performance when porting to di erent parallel platforms. Based on the LogP model, the proposed parameterized communication model can more accurately characterize the communication network of parallel platforms. The model encompasses a number of critical system parameters which can be easily measured on a given parallel platform. Based on the model, e cient methods to construct optimal multicast trees are proposed for both 1-port and -port communication architectures. Experimental results conducted on the IBM/SP at Argonne National Laboratory are presented to compare the performance of the optimal multicast tree with two other known tree-based multicast algorithms. We claim that our proposed multicast algorithms can be ported to di erent parallel platforms and provide a near-optimal performance as the truly machine-speci c optimal performance is achievable only when the underlying detailed network characteristics are considered.
2006 International Conference on Collaborative Computing: Networking, Applications and Worksharing, 2006
Sensors are envisioned to be at the center of distributed collaborative computing services involv... more Sensors are envisioned to be at the center of distributed collaborative computing services involving time-critical decision support. Sensors are small devices with limited communication and computational capabilities that collect data on their neighboring physical world and send the data periodically to server machines. Sensors form a collaborative network with these servers, where the sensors gather information and the servers perform various operations (e.g. filter, aggregate, join etc) on the information streams in real-time according to predefined queries or rules. Sensor data streams are continuous, un-ending and have highly volatile characteristics. As a result, traditional database systems are inappropriate for handling queries for sensor streams, and several stream data management systems have been proposed in the literature. In this paper we focus on a special type of query, namely join queries, which is the most expensive query operator. Here, we address the problem of finding an optimal join tree that maximizes throughput for sliding window based multi-join queries over continuous sensor data streams. We present a polynomial time algorithm Fodp and three variants of Fodp. Our experiments in ARES 1 show that for almost all instances, trees from Fodp and its variants perform close to the optimal trees from our exponential time algorithm OptDP [15], and significantly better than existing XJoin based heuristic algorithms.
Proceedings.Twenty-First Annual Joint Conference of the IEEE Computer and Communications Societies, 2002
Network survivability is a crucial requirement in high-speed optical networks. Typical approaches... more Network survivability is a crucial requirement in high-speed optical networks. Typical approaches of providing survivability have considered the failure of a single component such as a link or a node. In this paper, we consider a failure model in which any two links in the network may fail in an arbitrary order. Three loopback methods of recovering from double-link failures are presented. The first two methods require the identification of the failed links, while the third one does not. However, precomputing the backup paths for the third method is more difficult than for the first two.
Proceedings IEEE INFOCOM 2006. 25TH IEEE International Conference on Computer Communications, 2006
As the commercial success of the IEEE 802.11 protocol has made wireless infrastructure widely dep... more As the commercial success of the IEEE 802.11 protocol has made wireless infrastructure widely deployed, user organizations are increasingly concerned about the new vulnerabilities to their networks. While various security issues have been extensively studied, the threats posed by denial-of-service (DoS) attacks have not been fully exploited. In this paper, we consider DoS attacks posed by cheating on the backoff rules in the IEEE 802.11 DCF protocol and propose a scheme detecting such adversaries. Our scheme is based on the sequential hypothesis testing. We first develop analytical models for packet interarrival time distribution from each station in the network where multiple cheating stations co-exist. Using the characterization of this probability distribution, we develop an algorithm to detect cheating stations based on the throughout degradations observed at normal stations. Our simulation results show that the proposed algorithm only requires very small number of observations of packets with very small value (i.e., less than 0.1%) of false positive and false negative decisions. That is, our proposed algorithm performs significantly fast and also accurately.
With the increasing popularity of the World Wide Web (Web), it is very difficult for a single pop... more With the increasing popularity of the World Wide Web (Web), it is very difficult for a single popular Web server to handle the explosive demand from its clients. Web caching (proxy) is an effective technique to reduce a Web server's load, alleviate Internet traffic, and reduce user's response time when accessing a Web service. A fundamental question is where to
This paper considers the problem of allocating files in a document tree among multiple processors... more This paper considers the problem of allocating files in a document tree among multiple processors in a parallel webserver. It is assumed that access patterns are characterized by branching probabilities for an access that starts at a node and progresses down the tree. A combinatorial optimization problem is formulated that includes load balancing and communication costs. The general problem is
Our proposed scheme seeks to defend flooding- based distributed denial-of-service (DDoS) attacks ... more Our proposed scheme seeks to defend flooding- based distributed denial-of-service (DDoS) attacks in the Internet. An easy, yet very disruptive, way to cause unfairness to the legitimate users is to deplete the network bandwidth by sending high rate unresponsive flows from multiple sources. The network congestion created by such malicious flows causes most legitimate packets dropped at routers without reaching
This paper considers the problem of maximizing throughput in a multi-carrier wireless network tha... more This paper considers the problem of maximizing throughput in a multi-carrier wireless network that employs predictive link adaptation. The contributions of this paper are the following. We explicitly consider the time-penalty induced by the transmission-mode changes in wireless networks. We study the trade-offs between the channel-state prediction quality and the throughput. We provide sound offline problem analysis, with optimal or approximate solutions to the extent possible. We consider online version of the wireless channel scheduling problem, and provide a lower bound and propose algorithms that guarantee competitive ratios. We also extend our algorithms to accommodate the model when the actual channel capacity is different from as predicted earlier. Our results show that a modest consumption of resources for channel prediction and link adaptation may result in a significant throughput improvement, with only marginal gains through further enhancement of the prediction quality.
First International Conference on Broadband Networks, 2004
Channel-aware scheduling and link adaptation (LA) methods are widely considered to be crucial for... more Channel-aware scheduling and link adaptation (LA) methods are widely considered to be crucial for realizing high data rates in wireless networks. Multi-carrier systems that spread information bits over the entire signal band can take advantage of the frequency selective fading, and choose the sub-carrier(s) that have the best channel conditions for transmission. However, predicting the future channel states, and adjusting the transmission schedules and parameters accordingly, may consume valuable system resources, such as bandwidth, time, and power. Sub-carrier grouping, which refers to monitoring/treating a set of adjacent sub-carriers as a single unit, has been proposed to reduce the overhead associated with LA. This paper considers various models for sub-carrier grouping, each representing a different level of system complexity, and investigates the trade-offs between the sub-carrier grouping granularity and the link throughput. We first present an offline dynamic programming algorithm for finding an optimal solution, assuming perfect estimates of the channel states. The solution obtained through this formulation can serve as a performance bound for online algorithms. We then propose an online algorithm for the case of perfect channel state estimates, which is shown to have very close performance to the off-line optimal solution. This online algorithm is also extended to incorporate imperfect channel state estimates due to (i) variations in channel states among the sub-carriers within the same group, and (ii) prediction inaccuracies in estimating the future time-slots. Performances of the algorithms are evaluated through simulations and comparisons with the off-line solution. The trade-off between the sub-carrier grouping granularity and the link throughput is also presented. Overall, our results can provide some guidelines from the network scheduling perspective for deciding what kind of sub-carrier grouping model and granularity to use in multi-carrier networks with LA.
Channel-aware scheduling and link adaptation methods are widely considered to be crucial for real... more Channel-aware scheduling and link adaptation methods are widely considered to be crucial for realizing high data rates in wireless networks. However, predicting the future channel states, and adjusting the transmission schedules and parameters accordingly, may consume valuable system resources, such as bandwidth, time, and power. This paper considers the trade-offs between the prediction quality and the throughput in a wireless network that uses link adaptation and channel-aware scheduling. In particular, we study the effects of the look-ahead window, i.e., the range of future time slots on which we have channel state estimates, and the reliability of the channel state estimates on the throughput. We develop an online scheduling algorithm for a multi-channel multi-user network that employs predictive link adaptation, and generalize it to incorporate imperfect channel state estimates. We use this heuristic together with performance bounds to the offline version of the problem to evaluate the performance with varying prediction quality. Our results suggest that it may be possible to reap most of the potential channel-aware scheduling benefits with a small look-ahead, and imperfect channel state estimates. Thus, a modest consumption of resources for channel prediction and link adaptation may result in a significant throughput improvement, with only marginal gains through further enhancement of the prediction quality. Our results can provide meaningful guidelines in deciding what level of system resource consumption is justified for channel quality estimation and link adaptation.
2007 IEEE International Conference on Communications, 2007
CDMA 1x Ev-DO is an evolution of the CDMA2000 3G wireless standard to enable high rate packet dat... more CDMA 1x Ev-DO is an evolution of the CDMA2000 3G wireless standard to enable high rate packet data services up to 2.4Mbps in Rev 0, 3.1Mbps in Rev A, and 4.9Mbps in Rev B. To support such high data rates, CDMA 1x Ev-DO utilizes TDMA technology on the downlink allowing a single receiver per time slot. Accordingly, a scheduling algorithm is necessary to determine which user receives data in a given time slot. Typically, schedulers consider efficiency (increasing network throughput) or fairness (fair allocation of resources among users) as the basis for scheduling. To realize a reasonable tradeoff between efficiency and fairness, opportunistic schedulers have been proposed that take advantage of instantaneous improvements in users radio conditions by serving them at higher rates. However, given the increasing demand to support user QoS requirements, a QoS-aware scheduler may need to serve users even at times of inadequate radio conditions. In this paper, we propose a QoS-aware packet scheduling algorithm that takes into account policy rules that govern the relationships between different user QoS classes. Our scheduler uses marginal utility functions defined to embody the given rules when selecting users to be served on the downlink. Our scheduling algorithm is implemented in the OPNET module of the CDMA 1x Ev-DO system that we have developed, and its performance is compared to other scheduling algorithms. The simulation results show that our scheduler performs well ensuring the policy rules are followed, meeting users' QoS requirements, and providing fairness among users within the same QoS class.
2007 Fourth International Conference on Broadband Communications, Networks and Systems (BROADNETS '07), 2007
This paper considers the problem of maximizing the system throughput in IEEE 802.16 broadband acc... more This paper considers the problem of maximizing the system throughput in IEEE 802.16 broadband access networks with mesh topology, and the following results are presented. We first consider a simplified linear network with only uplink traffic and provide an optimal scheduling algorithm and establish an analytical result on the length of the schedule. We then consider the problem of routing and packet scheduling in general topology, and show its NPcompleteness. We also provide an ILP formulation for this problem. Based on our optimal algorithm for linear networks, we propose algorithms that find routes and schedules of packet transmissions in general mesh topologies. The performance of our proposed algorithms is analyzed using the NS-2 simulator. The results show that the suggested algorithms perform significantly better than other existing algorithms.
This paper studies a model for matching and scheduling a task structure onto a heterogeneous suit... more This paper studies a model for matching and scheduling a task structure onto a heterogeneous suite of computers. The paper first generalizes Optimal selection theory (OST) and its later augmentations to include general dependency graphs, which represent dependency between subtasks, and presents the Generalized optimal selection theory (GOST) model. The GOST model allows non-optimal choices of machines, as in AOST, heterogeneous code blocks, as in HOST, and incorporates communication time, system reconfiguration time, and data reformatting time (needed when data is exchanged between heterogeneous processors). Under the GOST model, we address the problem of matching/scheduling dependency graphs on a heterogeneous suite of computers. For arbitrary dependency graphs, matching/scheduling is NP-complete. But for the special instances of trees, and, more import.ant,ly, in the case of series-parallel dependency graphs we present polynomial time algorithms for optimal matching/scheduling . a 0-8186-5592-5/94 $3.00 0 1994 IEEE
Uploads
Papers by Hyeong-ah Choi