Paulo Jesus

A Survey of Distributed Data Aggregation Algorithms

Distributed data aggregation is an important task, allowing the decentralized determination of meaningful global properties, which can then be used to direct the execution of other applications. The resulting values are derived by the distributed computation of functions like Count, Sum, and Average. Some application examples deal with the determination of the network size, total storage capacity, average load, majorities and many others. In the last decade, many different approaches have been proposed, with different trade-offs in terms of accuracy, reliability, message and time complexity. Due to the considerable amount and variety of aggregation algorithms, it can be difficult and time consuming to determine which techniques will be more appropriate to use in specific settings, justifying the existence of a survey to aid in this task. This work reviews the state of the art on distributed data aggregation algorithms, providing three main contributions. First, it formally defines the concept of aggregation, characterizing the different types of aggregation functions. Second, it succinctly describes the main aggregation techniques, organizing them in a taxonomy. Finally, it provides some guidelines toward the selection and use of the most relevant techniques, summarizing their principal characteristics.

Extrema Propagation: Fast Distributed Estimation of Sums and Network Sizes

Aggregation of data values plays an important role on distributed computations, in particular, over peer-to-peer and sensor networks, as it can provide a summary of some global system property and direct the actions of self-adaptive distributed algorithms. Examples include using estimates of the network size to dimension distributed hash tables or estimates of the average system load to direct load balancing. Distributed aggregation using nonidempotent functions, like sums, is not trivial as it is not easy to prevent a given value from being accounted for multiple times; this is especially the case if no centralized algorithms or global identifiers can be used. This paper introduces Extrema Propagation, a probabilistic technique for distributed estimation of the sum of positive real numbers. The technique relies on the exchange of duplicate insensitive messages and can be applied in flood and/or epidemic settings, where multipath routing occurs; it is tolerant of message loss; it is fast, as the number of message exchange steps can be made just slightly above the theoretical minimum; and it is fully distributed, with no single point of failure and the result produced at every node.

Fault-Tolerant Aggregation by Flow Updating

Data aggregation plays an important role in the design of scalable systems, allowing the determination of meaningful system-wide properties to direct the execution of distributed applications. In the particular case of wireless sensor networks, data collection is often only practicable if aggregation is performed. Several aggregation algorithms have been proposed in the last few years, exhibiting different properties in terms of accuracy, speed and communication tradeoffs. Nonetheless, existing approaches are found lacking in terms of fault tolerance. In this paper, we introduce a novel fault-tolerant averaging based data aggregation algorithm. It tolerates substantial message loss (link failures), while competing algorithms in the same class can be affected by a single lost message. The algorithm is based on manipulating flows (in the graph theoretical sense), that are updated using idempotent messages, providing it with unique robustness capabilities. Furthermore, evaluation results obtained by comparing it with other averaging approaches have revealed that it outperforms them in terms of time and message complexity.

Flow updating

Data aggregation is a fundamental building block of modern distributed systems. Averaging based approaches, commonly designated gossip-based, are an important class of aggregation algorithms as they allow all nodes to produce a result, converge to any required accuracy, and work independently from the network topology. However, existing approaches exhibit many dependability issues when used in faulty and dynamic environments. This paper describes and evaluates a fault tolerant distributed aggregation technique, Flow Updating, which overcomes the problems in previous averaging approaches and is able to operate on faulty dynamic networks. Experimental results show that this novel approach outperforms previous averaging algorithms; it self-adapts to churn and input value changes without requiring any periodic restart, supporting node crashes and high levels of message loss, and works in asynchronous networks. Realistic concerns have been taken into account in evaluating Flow Updating, like the use of unreliable failure detectors and asynchrony, targeting its application to realistic environments. We describe a fault-tolerant data aggregation algorithm for dynamic networks.Experimental results show it outperforms previous averaging techniques.It self-adapts to churn and input value changes.It supports node crashes and high levels of message loss.It works in asynchronous settings.

Fault-Tolerant aggregation: flow-updating meets mass-distribution

Flow-Updating (FU) is a fault-tolerant technique that has proved to be efficient in practice for the distributed computation of aggregate functions in communication networks where individual processors do not have access to global information. Previous distributed aggregation protocols, based on repeated sharing of input values (or mass) among processors, sometimes called Mass-Distribution (MD) protocols, are not resilient to communication failures (or message loss) because such failures yield a loss of mass. In this paper, we present a protocol which we call Mass-Distribution with Flow-Updating (MDFU). We obtain MDFU by applying FU techniques to classic MD. We analyze the convergence time of MDFU showing that stochastic message loss produces low overhead. This is the first convergence proof of an FU-based algorithm. We evaluate MDFU experimentally, comparing it with previous MD and FU protocols, and verifying the behavior predicted by the analysis. Finally, given that MDFU incurs a fixed deviation proportional to the message-loss rate, we adjust the accuracy of MDFU heuristically in a new protocol called MDFU with Linear Prediction (MDFU-LP). The evaluation shows that both MDFU and MDFU-LP behave very well in practice, even under high rates of message loss and even changing the input values dynamically.

Spectra: robust estimation of distribution functions in networks

The distributed aggregation of simple aggregates such as minima/maxima, counts, sums and averages have been studied in the past and are important tools for distributed algorithms and network coordination. Nonetheless, this kind of aggregates may not be comprehensive enough to characterize biased data distributions or when in presence of outliers, making the case for richer estimates. This work presents Spectra, a distributed algorithm for the estimation of distribution functions over large scale networks. The estimate is available at all nodes and the technique depicts important properties: robustness when exposed to high levels of message loss, fast convergence speed and fine precision in the estimate. It can also dynamically cope with changes of the sampled local property and with churn, without requiring restarts. The proposed approach is experimentally evaluated and contrasted to a competing state of the art distribution aggregation technique.

Fault-tolerant aggregation: Flow-Updating meets Mass-Distribution

Flow-Updating (FU) is a fault-tolerant technique that has proved to be efficient in practice for the distributed computation of aggregate functions in communication networks where individual processors do not have access to global information. Previous distributed aggregation protocols, based on repeated sharing of input values (or mass) among processors, sometimes called Mass-Distribution (MD) protocols, are not resilient to communication failures (or message loss) because such failures yield a loss of mass. In this paper, we present a protocol which we call Mass-Distribution with Flow-Updating (MDFU). We obtain MDFU by applying FU techniques to classic MD. We analyze the convergence time of MDFU showing that stochastic message loss produces low overhead. This is the first convergence proof of an FU-based algorithm. We evaluate MDFU experimentally, comparing it with previous MD and FU protocols, and verifying the behavior predicted by the analysis. Finally, given that MDFU incurs a fixed deviation proportional to the message-loss rate, we adjust the accuracy of MDFU heuristically in a new protocol called MDFU with Linear Prediction (MDFU-LP). The evaluation shows that both MDFU and MDFU-LP behave very well in practice, even under high rates of message loss and even changing the input values dynamically.