We consider a system where randomly generated updates are to be transmitted monitor, but only single update can in the at time. Therefore, source has prioritize between two possible transmission policies: preempting current or discarding new one. Poisson arrivals and general service time, refer this as M/G/1/1 queue. start by studying average status age optimal arrival rate for these schemes under time distribution. then apply results on practical scenarios which sent through an erasure...

We study how coding in distributed storage reduces expected download time, addition to providing reliability against disk failures. The time is reduced because when a content file encoded with redundancy and across multiple disks, reading only subset of the disks sufficient for reconstruction. For same total used, exploits diversity better than simple replication, hence gives faster download. use novel fork-join queueing framework model users requesting simultaneously, derive bounds on time....

Network coding is an elegant and novel technique introduced at the turn of millennium to improve network throughput performance.It expected be a critical technology for networks future.This tutorial addresses first most natural questions one would ask about this new technique: how works what are its benefits, codes designed much it costs deploy implementing such codes, finally, whether there methods deal with cycles delay that present in all real networks.A companion issue deals primarily...

We propose a method to identify structural properties of multicast network configurations, by decomposing networks into regions through which the same information flows. This decomposition allows us show that very different are equivalent from coding point view, and offers means such equivalence classes. It also divide problem two almost independent tasks: one graph theory other classical channel theory. approach enables derive smallest code alphabet size sufficient any configuration with...

We consider a cache refresh system where local server is connected to multiple remote sources and maintains copies of the data items at sources. The each source updated randomly independently without notifying server, while refreshes corresponding cached periodically. freshness measured by two different metrics, <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">age synchronization</i> (AoS) information</i> (AoI). address following problem: given...

Using an age of information (AoI) metric, we examine the transmission coded updates through a binary erasure channel to monitor/receiver. We start by deriving average status update infinite incremental redundancy (IIR) system in which k-symbol continues until k symbols are received. This is then compared fixed (FR) each transmitted as n symbol packet and successfully received if only at least If fewer than received, discarded. Unlike IIR system, FR requires no feedback from receiver. For...

We consider the problem of securing a multicast network against wiretapper that can eavesdrop on packets limited number edges its choice. assume employs coding to simultaneously deliver available at source all destinations. show this be looked as generalization wiretap channel type II introduced in seminal paper by Ozarow and Wyner. In particular, we transmitted information secured using Ozarow–Wyner approach coset top existing code. This way, quickly transparently recover some results...

In cloud computing systems, assigning a task to multiple servers and waiting for the earliest copy finish is an effective method combat variability in response time of individual reduce latency. But adding redundancy may result higher cost resources, as well increase queueing delay due traffic load. This work helps understanding when how gives cost-efficient reduction For general service distribution, we compare different strategies terms number redundant tasks they are issued canceled. We...

Network coding is an elegant and novel technique introduced at the turn of millennium to improve network throughput performance.It expected be a critical technology for networks future.This tutorial deals with wireless content distribution networks, considered most likely applications coding, it also reviews emerging such as monitoring management.Multiple unicasts, security, unreliable links, quantum are addressed.The preceding companion theoretical foundations coding.

We evaluate the asymptotic normalized average distributions of a class combinatorial configurations in random, regular and irregular, binary low-density parity-check (LDPC) code ensembles. Among considered are trapping stopping sets. These sets represent subsets variable nodes Tanner graph that play an important role determining height point onset error-floor its performance curve. The techniques used for deriving spectra include large deviations theory statistical methods enumerating...

Color descriptors are among the most important features used in image analysis and retrieval. Due to its compact representation low complexity, direct histogram comparison is a commonly technique for measuring color similarity. However, it has many serious drawbacks, including high degree of dependency on codebook design, sensitivity quantization boundaries, inefficiency representing images with few dominant colors. In this paper, we present new algorithm matching that models behavior human...

We consider the problem of securing a multicast network against wiretapper that can intercept packets on limited number arbitrary links his choice. assume implements coding techniques to simultaneously deliver all available at source destinations. show how this be looked as generalization Ozarow-Wyner wiretap channel type II. In particular, we security achieved by using approach coset top implemented code. This way, quickly and transparently recover some results in literature secure for...

We study the fundamental trade-off between storage and content download time. show that time can be significantly reduced by dividing into chunks, encoding it to add redundancy then distributing across multiple disks. determine for two access models - fountain fork-join involve simultaneous access, individual from enqueued user requests respectively. For model we explicitly characterize time, while in derive upper lower bounds. Our results coding reduces through diversity of data more disks,...

We consider large-scale wireless sensor networks with n nodes, out of which k are in possession, (e.g., have sensed or collected some other way) information packets. In the scenarios network nodes vulnerable because of, for example, limited energy a hostile environment, it is desirable to disseminate acquired throughout so that each stores one (possibly coded) packet original source packets can be recovered, locally and computationally simple way from any k(1 + ¿) small ¿ > 0. develop...

To reduce computational complexity and delay in randomized network coded content distribution, for some other practical reasons, coding is not performed simultaneously over all blocks, but much smaller, possibly overlapping subsets of these known as generations. A penalty this strategy throughput reduction. analyze the loss, we model generations with random generation scheduling a coupon collector's brotherhood problem. This enables us to derive expected number packets needed successful...

Using age of information as the freshness metric, we examine a multicast network in which real-time status updates are generated by source and sent to group n interested receivers. We show that order keep at each receiver, should terminate transmission current update start sending new packet soon it receives acknowledgements back from any k out nodes. As stopping threshold increases, node is more likely get latest update, but most recent become outdated. derive minimized balances likelihood...

Hybrid ARQ transmission schemes combine the conventional with forward error correction. Incremental redundancy hybrid adapt their correcting code to varying channel gains, and thus achieve better throughput performance than ordinary ARQ, particularly over wireless channels fluctuating conditions. Consequently, scheme has been adopted by a number of standards for mobile phone networks. We provide brief survey theory state art present some possible future directions keeping in mind practical...

extended-abstract Share on Effective Straggler Mitigation: Which Clones Should Attack and When? Authors: Mehmet Fatih Aktas Rutgers University UniversityView Profile , Pei Peng Emina Soljanin Authors Info & Claims ACM SIGMETRICS Performance Evaluation ReviewVolume 45Issue 2Setember 2017 pp 12–14https://doi.org/10.1145/3152042.3152047Published:11 October 2017Publication History 28citation146DownloadsMetricsTotal Citations28Total Downloads146Last 12 Months31Last 6 weeks4 Get Citation AlertsNew...

We consider a multicast network in which real-time status updates generated by source are replicated and sent to multiple interested receiving nodes through independent links. The divided into two groups: one priority group consists of k that require the reception every update packet, other non-priority all without delivery requirement. Using age information as freshness metric, we analyze time-averaged at both nodes. For shifted-exponential link delay distributions, average node is lower...

We examine the throughput benefits that network coding offers with respect to average achievable by routing, where refers of rates individual receivers experience. relate these integrality gap a standard linear programming formulation for directed Steiner tree problem. describe families configurations over which at most doubles throughput, and analyze class graph N proportional /spl radic/N. also discuss other measures in networks, show how certain classes bounds can be translated into...

We consider large-scale networks with n nodes, out of which k are in possession, (e.g., have sensed or collected some other way) information packets. In the scenarios network nodes vulnerable because of, for example, limited energy a hostile environment, it is desirable to disseminate acquired throughout so that each stores one (possibly coded) packet and original source packets can be recovered later computationally simple way from any (1 + isin)k small isin > 0. developed two distributed...

short-paper Share on Queues with Redundancy: Latency-Cost Analysis Authors: Gauri Joshi MIT, Cambridge MA MAView Profile , Emina Soljanin Bell Labs Alcatel-Lucent, Murray Hill, NJ NJView Gregory Wornell Authors Info & Claims ACM SIGMETRICS Performance Evaluation ReviewVolume 43Issue 2September 2015 pp 54–56https://doi.org/10.1145/2825236.2825258Published:16 September 2015Publication History 30citation199DownloadsMetricsTotal Citations30Total Downloads199Last 12 Months17Last 6 weeks1 Get...

In cloud computing systems, assigning a job to multiple servers and waiting for the earliest copy finish is an effective method combat variability in response time of individual servers. Although adding redundant replicas always reduces service time, total spent per may be higher, thus increasing queue. The also proportional cost resources. We analyze how different redundancy strategies, eg. number replicas, when they are issued canceled, affect latency cost. get insight that log-concavity...