Y. Thomas Hou

Topology control for wireless sensor networks

We consider a two-tiered Wireless Sensor Network (WSN) consisting of sensor clusters deployed around strategic locations and base-stations (BSs) whose locations are relatively flexible. Within a sensor cluster, there are many small sensor nodes (SNs) that capture, encode and transmit relevant information from the designated area, and there is at least one application node (AN) that receives raw data from these SNs, creates a comprehensive local-view, and forwards the composite bit-stream toward a BS. In practice, both SN and AN are battery-powered and energy-constrained, and their node lifetimes directly affect the network lifetime of WSNs. In this paper, we focus on the topology control process for ANs and BSs, which constitute the upper tier of a two-tiered WSN. We propose approaches to maximize the topological network lifetime of the WSN, by arranging BS location and inter-AN relaying optimally. Based on an algorithm in Computational Geometry, we derive the optimal BS locations under three topological lifetime definitions according to mission criticality. In addition, by studying the intrinsic properties of WSNs, we establish the upper and lower bounds of their maximal topological lifetime. When inter-AN relaying becomes feasible and favorable, we continue to develop an optimal parallel relay allocation to further prolong the topological lifetime of the WSN. An equivalent serialized relay schedule is also obtained, so that each AN only needs to have one relay destination at any time throughout the mission. The experimental performance evaluation demonstrates the efficacy of topology control as a vital process to maximize the network lifetime of WSNs.

LRED: a robust active queue management scheme based on packet loss ratio

Active queue management (AQM) is an effective method to enhance congestion control, and to achieve tradeoff between link utilization and delay. The de facto standard, random early detection (RED), and most of its variants use queue length as a congestion indicator to trigger packet dropping. The proportional-integral (PI), use both queue length and traffic input rate as congestion indicators; effective stability model and practical design rules built on the TCP control model and abstracted AQM model reveal that such schemes enhance the stability of a system. In this paper, we propose an AQM scheme with fast response time, yet good robustness. The scheme, called loss ratio based RED (LRED), measures the latest packet loss ratio, and uses it as a complement to queue length in order to dynamically adjust packet drop probability. Employing the closed-form relationship between packet loss ratio and the number of TCP flows, this scheme is responsive even if the number of TCP flows varies significantly. We also provide the design rules for this scheme based on the well-known TCP control model. This schemes performance is examined under various network configurations, and compared to existing AQM schemes, including PI, random exponentially marking (REM), and adaptive virtual queue (AVQ). Our simulation results show that, with comparable complexity, this scheme has short response time, better robustness, and more desirable tradeoff than PI, REM, and AQV, especially under highly dynamic network and heavy traffic load.

An overview of DNS-based server selections in content distribution networks

With the explosive growth of the Internet, many new online services and applications are emerging. Some popular applications impose new challenges to the traditional Internet architecture and protocols. To alleviate the scalability burden in delivering popular Internet services to a large number of users, Web caching and content distribution technologies have been proposed, developed and deployed. Both approaches are designed to bring Web content closer to users and to improve their perceived quality of online experience. This paper surveys content distribution networks (CDNs), and in particular, their domain name system (DNS)-based server selection schemes. To bridge the gap between underlying principles and current practices, we choose a commercial content delivery provider, Akamai, as our focal case study. We first unveil Akamais content delivery network, as well as its site and object delivery technologies. We then examine the DNS-based server selection schemes and their variants in detail. Moreover, we offer some performance insights and discussions on their built-in strengths and weaknesses that are also applicable to other CDN providers.

A stable rate-based algorithm for active queue management

This paper proposes a rate-based active queue management algorithm or RAQM. It uses the aggregated traffic input rate to calculate packet drop probability according to an exponential rule. The proposed RAQM algorithm works in two modes: queue-independent and queue-dependent. In the queue-independent mode, it only relies on the aggregate traffic input rate to regulate the input rate to the expected value, i.e. expected link utility. In the queue-dependent mode, it also uses the instantaneous queue length to further adjust the packet drop probability and to regulate the queue length to the expected value. We analyze the stability and investigate practical implementation issues of the RAQM. Simulations are carried out to study RAQM performance and to compare with other AQM algorithms, in particular PI and REM schemes. The results demonstrate that RAQM achieves better stability and faster response as it can quickly regulate the queue length to the expected value with small overshoot. RAQM also obtains better tradeoff between link utilization and queuing delay, and obtains higher goodput with the same buffer size as in PI and REM schemes. Finally RAQM has O(1) complexity, thus independent of the number of flows.

On network bandwidth allocation policies and feedback control algorithms for packet networks

Abstract This paper summarizes our experience on the design of network bandwidth allocation policies and distributed rate calculation algorithms for packet-switched networks. In particular, we discuss two rate allocation policies: the generalized max–min (GMM) and the weight-proportional max–min (WPMM) policies, both of which generalize the classical max–min rate allocation policy. For the design of distributed algorithms to achieve these two rate allocation policies, we focus on rate-based distributed flow control where special control packets are employed to achieve the information exchange between a source and the network. We categorize two broad classes of distributed rate calculation algorithms in the literature using live algorithms as illustrations. To give insight, we compare the design tradeoffs between these two classes of algorithms in terms of performance objectives and implementation complexities. Furthermore, we discuss important extensions within each class of algorithms.

Rate allocation policies and explicit feedback control algorithms for packet networks

This paper presents an in-depth survey on network bandwidth allocation policies and discuss design methodologies of distributed rate calculation algorithms in packet-switched networks. In particular, we discuss two rate allocation policies: the generalized max-min and the generic weight- proportional max-min policies, both of which generalize the classical max-min rate allocation policy. For the design of distributed algorithms to achieve these two rate allocation policies, we focus on rate-based distributed flow control where special control packets are employed to achieve the information exchange between a source and the network. We categorize two broad classes of distributed rate calculation algorithms in the literature using live algorithms as illustrations. We compare the design tradeoffs between these two classes of algorithms in terms of performance objectives and implementation complexities and discuss important extensions within each class of algorithms.

Serialized optimal relay schedules in two-tiered wireless sensor networks

In two-tiered wireless sensor networks (WSNs), sensor nodes (SNs) are scattered in clusters, and are responsible for collecting relevant information from designated areas and transmitting to an application node (AN) in the cluster. The AN then constructs a local-view for the cluster by exploring correlations among information received from nearby SNs, and sends the local-view toward a base-station that creates a global-view for the entire WSN. ANs can also relay local-views for other ANs, if the resultant network lifetime is longer. In this paper, we want to arrange inter-AN relaying optimally, which is an important process in topology control for maximizing the topological lifetime of a WSN with regard to a certain amount of initial energy provisioning. We first propose some criteria on relay candidates preselection, which can considerably reduce the overhead of obtaining an optimal relaying. We then design an algorithm to serialize the parallel relay allocation, so that each AN only needs to have one relaying AN at any time. Finally, we demonstrate the equivalency in network lifetime of the serialized inter-AN relay schedules.

Retrieval and freshness thresholds in hierarchical caching systems

To scale up with the explosive Web growth, caching systems have been proposed and deployed over the Internet in recent years. Among them, hierarchical caching systems employing expiration-based consistency control mechanisms have become a viable and efficient solution. In this paper, we first analyze the performance of such hierarchical caching systems from the perspectives of both cache servers and end users. Then, we examine retrieval and freshness threshold-based approaches and their impact on system performance and user-perceived QoS. We show that by setting these thresholds appropriately, it is possible that (1) users can impose a consistency QoS requirement on the object that they wish to obtain without too much trade-off in system performance, and (2) performance bias against leaf users due to their unfavorable locations in the hierarchical structure can be mitigated.

Analysis and evaluation of expiration-based hierarchical caching systems

This paper investigates some fundamental properties and performance issues of the expiration-based caching systems. We focus on the hierarchical caching systems based on the time-to-live (TTL) expiration mechanism, and present a basic model for such systems. By analyzing the intrinsic timing behavior in this model, we derive some important performance metrics from the perspectives of caching systems and end users, respectively. We use network simulation results to further substantiate the efficacy of our analysis. Our results show some basic properties and trade-offs for a hierarchical caching system based on the weak consistency mechanism.