Kaliappa Ravindran

Multi-hop probing asymptotics in available bandwidth estimation: stochastic analysis

This paper analyzes the asymptotic behavior of packet-train probing over a multi-hop network path P carrying arbitrarily routed bursty cross-traffic flows. We examine the statistical mean of the packet-train output dispersions and its relationship to the input dispersion. We call this relationship the response curve of path P. We show that the real response curve Z is tightly lower-bounded by its multi-hop fluid counterpart F, obtained when every cross-traffic flow on P is hypothetically replaced with a constant-rate fluid flow of the same average intensity and routing pattern. The real curve Z asymptotically approaches its fluid counterpart F as probing packet size or packet train length increases. Most existing measurement techniques are based upon the single-hop fluid curve S associated with the bottleneck link in P. We note that the curve S coincides with F in a certain large-dispersion input range, but falls below F in the remaining small-dispersion input ranges. As an implication of these findings, we show that bursty cross-traffic in multi-hop paths causes negative bias (asymptotic underestimation) to most existing techniques. This bias can be mitigated by reducing the deviation of Z from S using large packet size or long packet-trains. However, the bias is not completely removable for the techniques that use the portion of S that falls below F.

A stochastic foundation of available bandwidth estimation: multi-hop analysis

This paper analyzes the asymptotic behavior of packet-train probing over a multi-hop network path P carrying arbitrarily routed bursty cross-traffic flows. We examine the statistical mean of the packet-train output dispersions and its relationship to the input dispersion. We call this relationship the response curve of path P. We show that the real response curve Z is tightly lower-bounded by its multi-hop fluid counterpart F, obtained when every cross-traffic flow on P is hypothetically replaced with a constant-rate fluid flow of the same average intensity and routing pattern. The real curve Z asymptotically approaches its fluid counterpart F as probing packet size or packet train length increases. Most existing measurement techniques are based upon the single-hop fluid curve S associated with the bottleneck link in P. We note that the curve S coincides with F in a certain large-dispersion input range, but falls below F in the remaining small-dispersion input ranges. As an implication of these findings, we show that bursty cross-traffic in multi-hop paths causes negative bias (asymptotic underestimation) to most existing techniques. This bias can be mitigated by reducing the deviation of Z from S using large packet size or long packet-trains. However, the bias is not completely removable for the techniques that use the portion of S that falls below F.

A queueing-theoretic foundation of available bandwidth estimation: single-hop analysis

Most existing available-bandwidth measurement techniques are justified using a constant-rate fluid cross-traffic model. To achieve a better understanding of the performance of current bandwidth measurement techniques in general traffic conditions, this paper presents a queueing-theoretic foundation of single-hop packet-train bandwidth estimation under bursty arrivals of discrete cross-traffic packets. We analyze the statistical mean of the packet-train output dispersion and its mathematical relationship to the input dispersion, which we call the probing-response curve. This analysis allows us to prove that the single-hop response curve in bursty cross-traffic deviates from that obtained under fluid cross traffic of the same average intensity and to demonstrate that this may lead to significant measurement bias in certain estimation techniques based on fluid models. We conclude the paper by showing, both analytically and experimentally, that the response-curve deviation vanishes as the packet-train length or probing packet size increases, where the vanishing rate is decided by the burstiness of cross-traffic.

What signals do packet-pair dispersions carry?

Although packet-pair probing has been used as one of the primary mechanisms to measure bottleneck capacity, cross-traffic intensity, and available bandwidth of end-to-end Internet paths, there is still no conclusive answer as to what information about the path is contained in the output packet-pair dispersions and how it is encoded. In this paper, we address this issue by deriving closed-form expression of packet-pair dispersion in the context of a single-hop path and general bursty cross-traffic arrival. Under the assumptions of cross-traffic stationarity and ASTA sampling, we examine the statistical properties of the information encoded in inter-packet spacings and derive the asymptotic average of the output packet-pair dispersions as a closed-form function of the input dispersion. We show that this result is different from what was obtained in prior work using fluid cross-traffic models and that this discrepancy has a significant impact on the accuracy of packet-pair bandwidth estimation.

Optimization algorithms for proxy server placement in content distribution networks

Popular Web sites receive an enormous share of Internet traffic. These sites have a competitive motivation to offer better service to their clients at lower cost. One of the solutions is to using content distribution network (CDN). When we design a CDN we need to find proxy server placement to provide its clients with the best available performance while consuming as little resource as possible. This is an optimization problem. Among the solutions greedy algorithm yields better result with low computational cost. The drawback is that it is easy to trap in the local optimum. We propose genetic algorithm to solve this problem. We mathematically model the optimization problem and then give details about how to apply genetic algorithm to proxy server placement problem. Simulation results for a simple topology are presented for both greedy algorithm and genetic algorithm.

Managing Robustness of Distributed Applications Under Uncertainties: An Information Assurance Perspective

The paper deals with how to manage the dependability of a distributed information system (DIS), in the presence of inaccuracies and partial knowledge of system models pertaining to failures of infrastructure components. The computational parts of a DIS are faced with two types of uncertainties: at ’information level’ due to the multi-dimensional nature of measurement data emanating from the external environment (i.e., the physical world), and at ’control level’ due to the incompleteness in knowledge about the application model for processing the measured data. These uncertainties have a compounded effect on the robustness of decisionmaking in the applications running on a DIS. Based on service-layer abstractions, the paper identifies applicationoriented metrics to quantify the quality of information flowing through a DIS. Suitable high-level quality metrics enable reasonably accurate decision-making by the application-layer control algorithms. Our approach is useful in the probabilistic management of: i) peer-to-peer communication assurance, and ii) fault-tolerance against component failures.

Model-based design of cyber-physical software systems for smart worlds: a software engineering perspective

The paper discusses the design of cyber-physical systems software around intelligent physical worlds (IPW). An IPW is the embodiment of control software functions wrapped around the external world processes, exhibiting self-adaptive behavior over a limited operating region of the system. This is in contrast with the traditional models where the physical world is basically dumb. A self-adaptation of IPW is feasible when certain system properties hold: function separability and piece-wise linearity of system behavioral models. The IPW interacts with an intelligent computational world (ICW) to work over wide range of operating conditions, by patching itself with suitable control parameters and rules & procedures relevant to a changed condition. The modular decomposition of a complex adaptive system into IPW and ICW has many advantages: lowering overall software complexity, simplifying system verification, and supporting easier evolution of system features. The paper illuminates our concept of IPW with software engineering-oriented case study of an industrial application: automotive system.

Adapting distributed voting algorithms for secure real-time embedded systems

Information assurance in real-time application settings requires dealing with extreme failure behaviors at the infrastructure level such as data corruptions by malicious processes and message timeliness violations in the network. Functional replication is employed to deal with such failures, with voting among the replica nodes to move correct pieces of data through the application-level subsystems. The goal is to develop a voting machinery that dynamically adjusts its internal mechanisms to deal with various types of failures. We present the design issues, with considerations of protocol correctness and performance engineering. A goal is to reduce the message overhead - and hence the power drain on wireless connected processes. The protocol is highly adaptive to deal with various types of failures occurring at the infrastructure level, in meeting the goal.

Software cybernetics to manage adaptation behavior of complex network systems

We employ model-based software (MBS) techniques to assess the quality of adaptation in a network system S in the presence of uncontrollable external environment conditions. The lack of complete knowledge about the I/O behavior of S arises due to large dimensionality of the input parameter space and their interactions with the various components in S. The MBS techniques adapt the operations of S over multiple “observe-adapt” rounds and steer S towards a reference input specs. How close is the actual behavior to the reference input depicts the quality of adaptation by S. The paper provides a cyber-physical systems (CPS) based software structure to evaluate the non-functional attributes of the output behavior of S, and autonomically adjust the network and algorithm parameters therein for effective control of S. The paper corroborates our CPS-based approach with a case study of QoE-aware video rate adaptation over a bandwidth-limited network path.

QoS Auditing for Evaluation of SLA in Cloud-based Distributed Services

Given cloud-based realization of a distributed system S, QoS auditing enables risk analysis and accounting of SLA violations under various security threats and resource depletions faced by S. The problem of QoS failures and security infringements arises due to the third-party control of cloud resources and components that are used in realizing the application-oriented service exported by S. The less-than-100% trust between the various sub-systems of S is a major issue that necessitates a probabilistic analysis of the application behavior relative to the SLA negotiated with S. In this light, QoS auditing allows reasoning about how good the SLA is complied by S in the face of hostile environment conditions. The paper describes case studies of CDN and replicated web service realized on a cloud.