arXiv subjects

Daniel E. Lucani

Publications and source records attributed to Daniel E. Lucani.

At least 37 records · Page 2Linked to original sources

arXiv2021

ZipLine: In-Network Compression at Line Speed

Network appliances continue to offer novel opportunities to offload processing from computing nodes directly into the data plane. One popular concern of network operators and their customers is to move data increasingly faster. A common technique to increase data throughput is to compress it before its transmission. However, this requires compression of the data -- a time and energy demanding pre-processing phase -- and decompression upon reception -- a similarly resource consuming operation. Moreover, if multiple nodes transfer similar data chunks across the network hop (e.g., a given pair of switches), each node effectively wastes resources by executing similar steps. This paper proposes ZipLine, an approach to design and implement (de)compression at line speed leveraging the Tofino hardware platform which is programmable using the P4_16 language. We report on lessons learned while building the system and show throughput, latency and compression measurements on synthetic and real-world traces, showcasing the benefits and trade-offs of our design.

cs.NI↗

arXiv2020

Optimal Accuracy-Time Trade-off for Deep Learning Services in Edge Computing Systems

With the increasing demand for computationally intensive services like deep learning tasks, emerging distributed computing platforms such as edge computing (EC) systems are becoming more popular. Edge computing systems have shown promising results in terms of latency reduction compared to the traditional cloud systems. However, their limited processing capacity imposes a trade-off between the potential latency reduction and the achieved accuracy in computationally-intensive services such as deep learning-based services. In this paper, we focus on finding the optimal accuracy-time trade-off for running deep learning services in a three-tier EC platform where several deep learning models with different accuracy levels are available. Specifically, we cast the problem as an Integer Linear Program, where optimal task scheduling decisions are made to maximize overall user satisfaction in terms of accuracy-time trade-off. We prove that our problem is NP-hard and then provide a polynomial constant-time greedy algorithm, called GUS, that is shown to attain near-optimal results. Finally, upon vetting our algorithmic solution through numerical experiments and comparison with a set of heuristics, we deploy it on a test-bed implemented to measure for real-world results. The results of both numerical analysis and real-world implementation show that GUS can outperform the baseline heuristics in terms of the average percentage of satisfied users by a factor of at least 50%.

cs.DC↗

arXiv2020

Yggdrasil: Privacy-aware Dual Deduplication in Multi Client Settings

This paper proposes Yggdrasil, a protocol for privacy-aware dual data deduplication in multi client settings. Yggdrasil is designed to reduce the cloud storage space while safeguarding the privacy of the client's outsourced data. Yggdrasil combines three innovative tools to achieve this goal. First, generalized deduplication, an emerging technique to reduce data footprint. Second, non-deterministic transformations that are described compactly and improve the degree of data compression in the Cloud (across users). Third, data preprocessing in the clients in the form of lightweight, privacy-driven transformations prior to upload. This guarantees that an honest-but-curious Cloud service trying to retrieve the client's actual data will face a high degree of uncertainty as to what the original data is. We provide a mathematical analysis of the measure of uncertainty as well as the compression potential of our protocol. Our experiments with a HDFS log data set shows that 49% overall compression can be achieved, with clients storing only 12% for privacy and the Cloud storing the rest. This is achieved while ensuring that each fragment uploaded to the Cloud would have 10^296 possible original strings from the client. Higher uncertainty is possible, with some reduction of compression potential.

cs.CR↗

arXiv2020

Memory-aware Online Compression of CAN Bus Data for Future Vehicular Systems

Vehicles generate a large amount of data from their internal sensors. This data is not only useful for a vehicle's proper operation, but it provides car manufacturers with the ability to optimize performance of individual vehicles and companies with fleets of vehicles (e.g., trucks, taxis, tractors) to optimize their operations to reduce fuel costs and plan repairs. This paper proposes algorithms to compress CAN bus data, specifically, packaged as MDF4 files. In particular, we propose lightweight, online and configurable compression algorithms that allow limited devices to choose the amount of RAM and Flash allocated to them. We show that our proposals can outperform LZW for the same RAM footprint, and can even deliver comparable or better performance to DEFLATE under the same RAM limitations.

cs.NI↗

arXiv2020

Hermes: Enabling Energy-efficient IoT Networks with Generalized Deduplication

With the advent of the Internet of Things (IoT), the ever growing number of connected devices observed in recent years and foreseen for the next decade suggests that more and more data will have to be transmitted over a network, before being processed and stored in data centers. Generalized deduplication (GD) is a novel technique to effectively reduce the data storage cost by identifying similar data chunks, and able to gradually reduce the pressure from the network infrastructure by limiting the data that needs to be transmitted. This paper presents Hermes, an application-level protocol for the data-plane that can operate over generalized deduplication, as well as over classic deduplication. Hermes significantly reduces the data transmission traffic while effectively decreasing the energy footprint, a relevant matter to consider in the context of IoT deployments. We fully implemented Hermes and evaluated its performance using consumer-grade IoT devices (e.g., Raspberry Pi 4B models). Our results highlight several trade-offs that must be taken into account when considering real-world workloads.

cs.DC↗

arXiv2019

Generalized Deduplication: Bounds, Convergence, and Asymptotic Properties

We study a generalization of deduplication, which enables lossless deduplication of highly similar data and show that standard deduplication with fixed chunk length is a special case. We provide bounds on the expected length of coded sequences for generalized deduplication and show that the coding has asymptotic near-entropy cost under the proposed source model. More importantly, we show that generalized deduplication allows for multiple orders of magnitude faster convergence than standard deduplication. This means that generalized deduplication can provide compression benefits much earlier than standard deduplication, which is key in practical systems. Numerical examples demonstrate our results, showing that our lower bounds are achievable, and illustrating the potential gain of using the generalization over standard deduplication. In fact, we show that even for a simple case of generalized deduplication, the gain in convergence speed is linear with the size of the data chunks.

cs.IT↗

arXiv2018

Reliable IoT Storage: Minimizing Bandwidth Use in Storage Without Newcomer Nodes

This letter characterizes the optimal policies for bandwidth use and storage for the problem of distributed storage in Internet of Things (IoT) scenarios, where lost nodes cannot be replaced by new nodes as is typically assumed in Data Center and Cloud scenarios. We develop an information flow model that captures the overall process of data transmission between IoT devices, from the initial preparation stage (generating redundancy from the original data) to the different repair stages with fewer and fewer devices. Our numerical results show that in a system with 10 nodes, the proposed optimal scheme can save as much as 10.3% of bandwidth use, and as much as 44% storage use with respect to the closest suboptimal approach.

cs.NI↗

arXiv2015

Analysis and Optimization of Sparse Random Linear Network Coding for Reliable Multicast Services

Point-to-multipoint communications are expected to play a pivotal role in next-generation networks. This paper refers to a cellular system transmitting layered multicast services to a multicast group of users. Reliability of communications is ensured via different Random Linear Network Coding (RLNC) techniques. We deal with a fundamental problem: the computational complexity of the RLNC decoder. The higher the number of decoding operations is, the more the user's computational overhead grows and, consequently, the faster the battery of mobile devices drains. By referring to several sparse RLNC techniques, and without any assumption on the implementation of the RLNC decoder in use, we provide an efficient way to characterize the performance of users targeted by ultra-reliable layered multicast services. The proposed modeling allows to efficiently derive the average number of coded packet transmissions needed to recover one or more service layers. We design a convex resource allocation framework that allows to minimize the complexity of the RLNC decoder by jointly optimizing the transmission parameters and the sparsity of the code. The designed optimization framework also ensures service guarantees to predetermined fractions of users. The performance of the proposed optimization framework is then investigated in a LTE-A eMBMS network multicasting H.264/SVC video services.

cs.IT↗

arXiv2015

Fulcrum Network Codes: A Code for Fluid Allocation of Complexity

This paper proposes Fulcrum network codes, a network coding framework that achieves three seemingly conflicting objectives: (i) to reduce the coding coefficient overhead to almost n bits per packet in a generation of n packets; (ii) to operate the network using only GF(2) operations at intermediate nodes if necessary, dramatically reducing complexity in the network; (iii) to deliver an end-to-end performance that is close to that of a high-field network coding system for high-end receivers while simultaneously catering to low-end receivers that decode in GF(2). As a consequence of (ii) and (iii), Fulcrum codes have a unique trait missing so far in the network coding literature: they provide the network with the flexibility to spread computational complexity over different devices depending on their current load, network conditions, or even energy targets in a decentralized way. At the core of our framework lies the idea of precoding at the sources using an expansion field GF(2h) to increase the number of dimensions seen by the network using a linear mapping. Fulcrum codes can use any high-field linear code for precoding, e.g., Reed-Solomon, with the structure of the precode determining some of the key features of the resulting code. For example, a systematic structure provides the ability to manage heterogeneous receivers while using the same data stream. Our analysis shows that the number of additional dimensions created during precoding controls the trade-off between delay, overhead, and complexity. Our implementation and measurements show that Fulcrum achieves similar decoding probability as high field Random Linear Network Coding (RLNC) approaches but with encoders/decoders that are an order of magnitude faster.

cs.IT↗

arXiv2012

Whether and Where to Code in the Wireless Relay Channel

The throughput benefits of random linear network codes have been studied extensively for wirelined and wireless erasure networks. It is often assumed that all nodes within a network perform coding operations. In energy-constrained systems, however, coding subgraphs should be chosen to control the number of coding nodes while maintaining throughput. In this paper, we explore the strategic use of network coding in the wireless packet erasure relay channel according to both throughput and energy metrics. In the relay channel, a single source communicates to a single sink through the aid of a half-duplex relay. The fluid flow model is used to describe the case where both the source and the relay are coding, and Markov chain models are proposed to describe packet evolution if only the source or only the relay is coding. In addition to transmission energy, we take into account coding and reception energies. We show that coding at the relay alone while operating in a rateless fashion is neither throughput nor energy efficient. Given a set of system parameters, our analysis determines the optimal amount of time the relay should participate in the transmission, and where coding should be performed.

cs.NI↗

arXiv2012

Systematic Network Coding with the Aid of a Full-Duplex Relay

A characterization of systematic network coding over multi-hop wireless networks is key towards understanding the trade-off between complexity and delay performance of networks that preserve the systematic structure. This paper studies the case of a relay channel, where the source's objective is to deliver a given number of data packets to a receiver with the aid of a relay. The source broadcasts to both the receiver and the relay using one frequency, while the relay uses another frequency for transmissions to the receiver, allowing for a full-duplex operation of the relay. We analyze the decoding complexity and delay performance of two types of relays: one that preserves the systematic structure of the code from the source; another that does not. A systematic relay forwards uncoded packets upon reception, but transmits coded packets to the receiver after receiving the first coded packet from the source. On the other hand, a non-systematic relay always transmits linear combinations of previously received packets. We compare the performance of these two alternatives by analytically characterizing the expected transmission completion time as well as the number of uncoded packets forwarded by the relay. Our numerical results show that, for a poor channel between the source and the receiver, preserving the systematic structure at the relay (i) allows a significant increase in the number of uncoded packets received by the receiver, thus reducing the decoding complexity, and (ii) preserves close to optimal delay performance.

cs.NI↗

arXiv2011

Energy-Delay Considerations in Coded Packet Flows

We consider a line of terminals which is connected by packet erasure channels and where random linear network coding is carried out at each node prior to transmission. In particular, we address an online approach in which each terminal has local information to be conveyed to the base station at the end of the line and provide a queueing theoretic analysis of this scenario. First, a genie-aided scenario is considered and the average delay and average transmission energy depending on the link erasure probabilities and the Poisson arrival rates at each node are analyzed. We then assume that all nodes cannot send and receive at the same time. The transmitting nodes in the network send coded data packets before stopping to wait for the receiving nodes to acknowledge the number of degrees of freedom, if any, that are required to decode correctly the information. We analyze this problem for an infinite queue size at the terminals and show that there is an optimal number of coded data packets at each node, in terms of average completion time or transmission energy, to be sent before stopping to listen.

cs.IT↗

arXiv2011

On the Order Optimality of Large-scale Underwater Networks

Capacity scaling laws are analyzed in an underwater acoustic network with $n$ regularly located nodes on a square, in which both bandwidth and received signal power can be limited significantly. A narrow-band model is assumed where the carrier frequency is allowed to scale as a function of $n$. In the network, we characterize an attenuation parameter that depends on the frequency scaling as well as the transmission distance. Cut-set upper bounds on the throughput scaling are then derived in both extended and dense networks having unit node density and unit area, respectively. It is first analyzed that under extended networks, the upper bound is inversely proportional to the attenuation parameter, thus resulting in a highly power-limited network. Interestingly, it is seen that the upper bound for extended networks is intrinsically related to the attenuation parameter but not the spreading factor. On the other hand, in dense networks, we show that there exists either a bandwidth or power limitation, or both, according to the path-loss attenuation regimes, thus yielding the upper bound that has three fundamentally different operating regimes. Furthermore, we describe an achievable scheme based on the simple nearest-neighbor multi-hop (MH) transmission. We show that under extended networks, the MH scheme is order-optimal for all the operating regimes. An achievability result is also presented in dense networks, where the operating regimes that guarantee the order optimality are identified. It thus turns out that frequency scaling is instrumental towards achieving the order optimality in the regimes. Finally, these scaling results are extended to a random network realization. As a result, vital information for fundamental limits of a variety of underwater network scenarios is provided by showing capacity scaling laws.

cs.IT↗

arXiv2010

On Capacity Scaling of Underwater Networks: An Information-Theoretic Perspective

Capacity scaling laws are analyzed in an underwater acoustic network with $n$ regularly located nodes on a square. A narrow-band model is assumed where the carrier frequency is allowed to scale as a function of $n$. In the network, we characterize an attenuation parameter that depends on the frequency scaling as well as the transmission distance. A cut-set upper bound on the throughput scaling is then derived in extended networks. Our result indicates that the upper bound is inversely proportional to the attenuation parameter, thus resulting in a highly power-limited network. Interestingly, it is seen that unlike the case of wireless radio networks, our upper bound is intrinsically related to the attenuation parameter but not the spreading factor. Furthermore, we describe an achievable scheme based on the simple nearest neighbor multi-hop (MH) transmission. It is shown under extended networks that the MH scheme is order-optimal as the attenuation parameter scales exponentially with $\sqrt{n}$ (or faster). Finally, these scaling results are extended to a random network realization.

cs.IT↗

arXiv2009

Capacity Scaling Laws for Underwater Networks

The underwater acoustic channel is characterized by a path loss that depends not only on the transmission distance, but also on the signal frequency. Signals transmitted from one user to another over a distance $l$ are subject to a power loss of $l^{-α}{a(f)}^{-l}$. Although a terrestrial radio channel can be modeled similarly, the underwater acoustic channel has different characteristics. The spreading factor $α$, related to the geometry of propagation, has values in the range $1 \leq α\leq 2$. The absorption coefficient $a(f)$ is a rapidly increasing function of frequency: it is three orders of magnitude greater at 100 kHz than at a few Hz. Existing results for capacity of wireless networks correspond to scenarios for which $a(f) = 1$, or a constant greater than one, and $α\geq 2$. These results cannot be applied to underwater acoustic networks in which the attenuation varies over the system bandwidth. We use a water-filling argument to assess the minimum transmission power and optimum transmission band as functions of the link distance and desired data rate, and study the capacity scaling laws under this model.

cs.IT↗

arXiv2009

Random Linear Network Coding for Time-Division Duplexing: Queueing Analysis

We study the performance of random linear network coding for time division duplexing channels with Poisson arrivals. We model the system as a bulk-service queue with variable bulk size. A full characterization for random linear network coding is provided for time division duplexing channels [1] by means of the moment generating function. We present numerical results for the mean number of packets in the queue and consider the effect of the range of allowable bulk sizes. We show that there exists an optimal choice of this range that minimizes the mean number of data packets in the queue.

cs.IT↗

arXiv2009

Broadcasting in Time-Division Duplexing: A Random Linear Network Coding Approach

We study random linear network coding for broadcasting in time division duplexing channels. We assume a packet erasure channel with nodes that cannot transmit and receive information simultaneously. The sender transmits coded data packets back-to-back before stopping to wait for the receivers to acknowledge the number of degrees of freedom, if any, that are required to decode correctly the information. We study the mean time to complete the transmission of a block of packets to all receivers. We also present a bound on the number of stops to wait for acknowledgement in order to complete transmission with probability at least $1-ε$, for any $ε>0$. We present analysis and numerical results showing that our scheme outperforms optimal scheduling policies for broadcast, in terms of the mean completion time. We provide a simple heuristic to compute the number of coded packets to be sent before stopping that achieves close to optimal performance with the advantage of a considerable reduction in the search time.

cs.IT↗

arXiv2009

Random Linear Network Coding For Time Division Duplexing: Energy Analysis

We study the energy performance of random linear network coding for time division duplexing channels. We assume a packet erasure channel with nodes that cannot transmit and receive information simultaneously. The sender transmits coded data packets back-to-back before stopping to wait for the receiver to acknowledge the number of degrees of freedom, if any, that are required to decode correctly the information. Our analysis shows that, in terms of mean energy consumed, there is an optimal number of coded data packets to send before stopping to listen. This number depends on the energy needed to transmit each coded packet and the acknowledgment (ACK), probabilities of packet and ACK erasure, and the number of degrees of freedom that the receiver requires to decode the data. We show that its energy performance is superior to that of a full-duplex system. We also study the performance of our scheme when the number of coded packets is chosen to minimize the mean time to complete transmission as in [1]. Energy performance under this optimization criterion is found to be close to optimal, thus providing a good trade-off between energy and time required to complete transmissions.

cs.IT↗