We present BeRGeR: the first asynchronous geometric routing algorithm that guarantees delivery of a message despite a Byzantine fault without relying on cryptographic primitives or randomization. The communication graph is a planar embedding that remains three-connected if all edges intersecting the source-target line segment are removed. We prove the algorithm correct and estimate its message complexity.
This work in progress paper outlines research looking at the performance impact of using different storage interfaces to access the high performance object store DAOS. We demonstrate that using DAOS through a FUSE based filesystem interface can provide high performance, but there are impacts when choosing what I/O library or interface to utilises, with HDF5 exhibiting the highest impact. However, this varied depending on what type of I/O operations were undertaken.
We propose a novel Quality of Experience (QoE) metric as a key criterion to optimize the composition of energy services in a crowdsourced IoT environment. A novel importance-based composition algorithm is proposed to ensure the highest QoE for consumers. A set of experiments is conducted to evaluate the proposed approaches' effectiveness and efficiency.
We describe a distributed, asynchronous variant of Edmonds's exact algorithm for producing perfect matchings of minimum weight. The development of this algorithm is driven by an application to online error correction in quantum computing, first envisioned by Fowler; we analyze the performance of our algorithm as applied to this domain in a sequel.
This paper studies the fundamental problem of data persistency for a general family of redundancy schemes in distributed storage systems, called replicated erasure codes. Namely, we analyze two strategies of replicated erasure codes distribution: random and symmetric. For both strategies we derive closed analytical and asymptotic formulas for expected data persistency despite nodes failure.
A method for evaluating the relative performance of local, cloud and combined processing of divisible (i.e. partitionable) data loads is presented. It is shown how to do this in the context of Amdahl's law. A single level (star) network operating under each of three fundamental scheduling policies is used as an example. Applications include mobile computing, cloud computing and signature searching.
Cloud providers are facing a complex problem in configuring software applications ready for deployment on their infrastructures. Hierarchical Task Network (HTN) planning can provide effective means to solve such deployment problems. We present an HTN planning domain that models deployment problems as found in realistic Cloud environments.
This paper describes a parallel implementation of Viterbi decoding algorithm. Viterbi decoder is widely used in many state-of-the-art wireless systems. The proposed solution optimizes both throughput and memory usage by applying optimizations such as unified kernel implementation and parallel traceback. Experimental evaluations show that the proposed solution achieves higher throughput compared to previous GPU-accelerated solutions.
Miquel Pericas, Oscar Palomar, Vassilis Papaefstathiou
et al.
This volume contains the proceedings of the 13th International Workshop on Programmability and Architectures for Heterogeneous Multicores. The workshop was held in conjunction with the 16th International Conference on High-Performance and Embedded Architectures and Compilers (HiPEAC) in Bologna, Italy on January 20th, 2020.
We follow existing distributed systems frameworks employing methods from algebraic topology to formally define primitives of blockchain technology. We define the notion of cross chain liquidity, sharding and probability spaces between and within blockchain protocols. We incorporate recent advancements in synthetic homology to show that this topological framework can be implemented within a type system. We use recursion schemes to define kernels admitting smooth manifolds across protocol complexes, leading to the formal definition of a Poincare protocol.
Game-theoretical approach to the analysis of parallel algorithms is proposed. The approach is based on presentation of the parallel computing as a congestion game. In the game processes compete for resources such as core of a central processing unit and a communication subsystem. There are players, resources and payoffs (time delays) of players which depend on resources usage. Comparative analysis of various optimality principles in the proposed model may be performed.
This paper presents the integration of Dijkstra's algorithm within a Blackboard framework to optimize the selection of web services from service providers. In addition, methods are presented how dynamic changes during the workflow execution can be handled; specifically, how changes of the service parameters have effects on the system. For justification of our approach, and to show practical feasibility, a sample implementation is presented.
Scientific workflow is a powerful tool to streamline and organize computational steps of scientific application. This paper presents Emerald, a system that adds sophisticated cloud offloading capabilities to scientific workflows. Emerald automatically offloads computation intensive steps of scientific workflow to the cloud in order to enhance workflow performance. Emerald provides easy-to-use APIs to help developers build cloud offloading enabled scientific workflows. Evaluation showed that Emerald can effectively reduce up to 55% of execution time for scientific applications.
We present a simple deterministic distributed $(2+ε)$-approximation algorithm for minimum weight vertex cover, which completes in $O(\logΔ/ε\log\logΔ)$ rounds, where $Δ$ is the maximum degree in the graph, for any $ε>0$ which is at most $O(1)$. For a constant $ε$, this implies a constant approximation in $O(\logΔ/\log\logΔ)$ rounds, which contradicts the lower bound of [KMW10].
This article shows a lower cost realization of a compute cluster using Debian distribution such as PelicanHPC. We will explain parameterization and network configuration for master and compute slave nodes. Performance testing will take place using flops.f file given by MPI. The results will be compared between differents clusters. We will explain quickly how the temperature is controlled by a microcontroller unit.
This paper indicates two errors in the formulation of the main optimization model in the article "Dynamic power management in energy-aware computer networks and data intensive computing systems" by Niewiadomska-Szynkiewicz et al. [FGCS, vol.37 (2014), pp.284-296] and shows how to fix them.
Modern educational institutions widely used virtual laboratories and cloud technologies. In practice must deal with security, processing speed and other tasks. The paper describes the experience of the construction of an experimental stand cloud computing and network management. Models and control principles set forth herein.