Ziqian Dong

Network measurement based modeling and optimization for IP geolocation

IP geolocation plays a critical role in location-aware network services and network security applications. Commercially deployed IP geolocation databases may provide outdated or incorrect location of Internet hosts due to slow record updates and dynamic IP address assignment by the ISPs. Measurement-based IP geolocation is used to provide real time location estimation of Internet hosts based on network delays. This paper proposes a measurement-based IP geolocation framework that provides location estimation of an Internet host in real time. The proposed frame work models the relationship between measured network delays and geographic distances using segmented polynomial regression model and semidefinite programming for optimization. Weighted and non-weighted schemes are evaluated for location estimation. The proposed framework shows close to 17 and 26 miles median estimation error for nodes in North America and Europe, respectively. The proposed schemes achieve 70-80% improvement in median estimation error comparing to the first order regression approach for experimental data collected from Planet-Lab.

Load-balanced combined input-crosspoint buffered packet switch and long round-trip times

The amount of memory in buffered crossbars is proportional to the number of crosspoints, or O(N/sup 2/), where N is the number of ports, and to the crosspoint buffer size, which is defined by the distance between the line cards and the buffered crossbar, to achieve 100% throughput under high-speed data flows. A long distance between these two components can make a buffered crossbar costly to implement. In this letter, we propose a load-balanced combined input-crosspoint buffered packet switch that uses small crosspoint buffers and no speedup. The proposed switch reduces the required size of the crosspoint buffers by a factor of N and keeps the cells in sequence.

Greedy scheduling of tasks with time constraints for energy-efficient cloud-computing data centers

In this paper, we introduce a model of task scheduling for a cloud-computing data center to analyze energy-efficient task scheduling. We formulate the assignments of tasks to servers as an integer-programming problem with the objective of minimizing the energy consumed by the servers of the data center. We prove that the use of a greedy task scheduler bounds the constraint service time whilst minimizing the number of active servers. As a practical approach, we propose the most-efficient-server-first task-scheduling scheme to minimize energy consumption of servers in a data center. Most-efficient-server-first schedules tasks to a minimum number of servers while keeping the data-center response time within a maximum constraint. We also prove the stability of most-efficient-server-first scheme for tasks with exponentially distributed, independent, and identically distributed arrivals. Simulation results show that the server energy consumption of the proposed most-efficient-server-first scheduling scheme is 70 times lower than that of a random-based task-scheduling scheme.

Schemes for Fast Transmission of Flows in Data Center Networks

In this paper, we survey different existing schemes for the transmission of flows in Data Center Networks (DCNs). The transport of flows in DCNs must cope with the bandwidth demands of the traffic that a large number of data center applications generates and achieve high utilization of the data center infrastructure to make the data center financially viable. Traffic in DCNs roughly comprises short flows, which are generated by the Partition/Aggregate model adopted by several applications and have sizes of a few kilobytes, and long flows, which are data for the operation and maintenance of the data center and have sizes on the order of megabytes. Short flows must be transmitted (or completed) as soon as possible or within a deadline, and long flows must be serviced with a minimum acceptable throughput. The coexistence of short and long flows may jeopardize achieving both performance objectives simultaneously. This challenge has motivated growing research on schemes for managing the transmission of flows in DCNs. We describe several recent schemes aimed at reducing the flow completion time in DCNs. We also present a summary of existing solutions for the incast traffic phenomenon. We provide a comparison and classification of the surveyed schemes, describe their advantages and disadvantages, and show the different trends for scheme design. For completeness, we describe some DCN architectures, discuss the traffic patterns of DCNs, and discuss why some existing versions of transport protocols may not be usable in DCNs. At the end, we discuss some of the identified research challenges.

Task Scheduling and Server Provisioning for Energy-Efficient Cloud-Computing Data Centers

In this paper, we present an optimization model for task scheduling for minimizing energy consumption in cloud-computing data centers. The proposed approach was formulated as an integer programming problem to minimize the cloud-computing data center energy consumption by scheduling tasks to a minimum numbers of servers while keeping the task response time constraints. We prove that the average task response time and the number of active servers needed to meet such time constraints are bounded through the use of a greedy task-scheduling scheme. In addition, we propose the most-efficient server- first task-scheduling scheme to minimize energy expenditure as a practical scheduling scheme. We model and simulate the proposed scheduling scheme for a data center with heterogeneous tasks. The simulation results show that the proposed task-scheduling scheme reduces server energy consumption on average over 70 times when compared to the energy consumed under a (not-optimized) random-based task-scheduling scheme. We show that energy savings are achieved by minimizing the allocated number of servers.

Non-blocking memory-memory-memory Clos-network packet switch

Buffered Clos-network switch, also referred as memory-memory-memory (MMM) Clos-network switch, is an alternative to single-stage switches to implement large-scale packet switches. In this paper, we unveil the head-of-line blocking problem of MMM Clos-network switches with switch modules implemented by buffered crossbars, which causes performance degradation. We propose a three-stage buffered Clos-network switch with per-output flow queues in the switch modules at the first two stages to avoid head-of-line blocking. The proposed switch, called the MMeM switch, achieves higher performance than an MMM buffered Clos-network packet switch. We show the performance improvement of the proposed MMeM switch under different traffic patterns.

Memory-memory-memory Clos-network packet switches with in-sequence service

Out-of-sequence is a problem faced by multi-stage buffered Clos-network switches. This paper proposes two buffered three-stage Clos-network packet switches that service packets in sequence and provide high switching performance. The proposed switches require short configuration times as compared to existing bufferless or partially buffered Clos-network switches. The proposed switches use time stamps assigned at the input modules to identify the order of packets in the switch. The switches use time-stamp monitoring mechanisms either at the input modules in a switch called the MMM-IM switch, or at the output modules in a switch called the MMM-OM switch to keep packets in sequence. Synchronization among different switch modules is not required in the proposed switches. The switching performance study presented in this paper shows that in-sequence monitoring at the IM provides higher performance and larger scalability than in-sequence monitoring at the output. Furthermore, the throughput of the MMM-IM switch is comparable to that of a switch that may service packets out of sequence.

Parallel Search Trie-Based Scheme for Fast IP Lookup

As data rates in the Internet increase, the Internet Protocol (IP) address lookup is required to be resolved in shorter resolution times. IP address lookup involves finding the longest matching prefix from a database of prefixes that better matches the destination address of a packet. The fastest IP-address lookup solutions are based on ternary content addressable memories (TCAMs), which can resolve the IP lookup in one memory-access time. However, TCAMs have a high power consumption and large complexity that may limit their scalability and storage capacity. An alternative is to use random access memory (RAM) that stores a forwarding table in a trie form. Proposed trie-based solutions for IP lookup require three or more memory-access times in the worst-case scenario. This makes them unattractive despite their reduced power consumption. In this paper, we propose a flexible and fast trie-based IP-lookup algorithm where parallel searching is performed. This algorithm performs lookup in two memory- access times whith a feasible amount of memory or three memory access times with reduced memory.

Indoor localization for mobile devices

This paper proposes an indoor localization system for mobile devices in urban high-rise environments. The proposed system classifies received signal strength measured from existing Wi-Fi access points, and predicts location of mobile devices based on the measured Wi-Fi signal strength and building floor plan. We collected data using different mobile devices, generated heat maps of signal strength recorded in a high-rise building for each Wi-Fi access point, and evaluated three location estimation methods. We applied clustering and Naive-Bayes algorithms to train the classifier and compared the location estimation accuracy of the three methods on the collected dataset. Experimental results show that the system can achieve an average of over 80% location prediction accuracy by clustering data into a number of location zones for the dataset.

Long round-trip time support with shared-memory crosspoint buffered packet switch

The amount of memory in buffered crossbars in combined input-crosspoint buffered switches is proportional to the number of crosspoints, or O(N/sup 2/), where N is the number of ports, and to the crosspoint buffer size, which is defined by the distance between the line cards and the buffered crossbar, to achieve 100% throughput under port-rate data flows. A long distance between these two components can make a buffered crossbar costly to implement. In this paper, we propose and examine two shared-memory crosspoint buffered packet switches that use small crosspoint buffers to support a long round-trip time, which is mainly affected by the transmission delay caused by the distance between line cards and the buffered crossbar. The proposed switch reduces the required buffer memory of the buffered crossbar by 50% or more. We show that a shared-memory crosspoint buffer switch can provide high this improvement without speedup.