This paper proposes the optimal protection scheme for the reliable distributed storage in the Internet of Things (IoT) scenarios, where the system loses the random number of devices per failure event. We investigate the protection problem with the random number of failures, and model the protection strategies based on a path graph. Furthermore, we devise an algorithm to identify the optimal strategy (OP) that achieves the minimum bandwidth use with low complexity. The numerical results show that in a system with 10 nodes, our proposed optimal strategy reduces the protection bandwidth up to more than five times the original data size, and the number of connections as much as 74 %, with respect to a strategy generalized from the case of one by one failure.
Next generation use-cases of wireless networks require a great deal of flexibility in order to adopt to the constantly changing state of innovation and to accommodate future application requirements. Header compression has been an ever present solution since the advent of wireless networks and the most current version of it, Robust Header Compression (RoHC), has seen a widespread adoption in Long Term Evolution (LTE) cellular networks. Recent research has mostly focused on the integration and enhancement of RoHC, instead of advancing the core concept of the compression. In this paper we present for the first time a novel design that can tackle the compression of arbitrary packet streams regardless of the employed protocols and the transmitted data. We present our initial findings for an error-free scenario with various simulated and real-life packet streams and show that even with the absence of design time knowledge about the packet structures, one can compress a significant part of the streams, yielding compression gains for IP packets up to 90 %, as an example.
Over the past few years, the Dynamic Adaptive Streaming over HTTP (DASH) standard has been widely adopted by video streaming services; this has led to a considerable amount of research on efficient adaptation algorithms to maximize users' Quality of Experience (QoE), but the comparison between algorithms is often flawed and unrealistic. Trace-based simulations or experiments in wired testbeds often fail to capture the full complexity and variability of a live network environment, and the comparison between them is often biased by the simplistic assumptions made about the simulation/testing setup. In this work, we implement four of the most representative adaptation algorithms and compare their performance in a real campus network. This allows us to have a fair comparison of the various algorithms' strengths and weaknesses in a representative scenario, improving the understanding of the dynamics of video streaming adaptation and highlighting the open problems in the field. The test results make a clear trade-off emerge, as algorithms designed for high fairness in a static scenario cannot deal with a dynamic channel and cross-traffic effectively and result in very low QoE levels, while more aggressive algorithms face rebuffering events in fast-varying scenarios.
We consider a two-way relay channel (TWRC), in which two source nodes exchange their data with the help of a relay node using network coding (NC). The sources, to be exchanged are double-layer bit-streams, can be partitioned into a high-priority (HP) bit-stream and a low-priority (LP) bit-stream. The HP bit-stream should be generally more protected than the LP bit-stream because it is assumed that the HP bit-stream includes more important data. Such an assumption is applicable to video communications using a base and enhancement layers of scalable video coding (SVC). In this paper, we propose an NC mapping rule that can continuously improve bit error rate (BER) performance even when the relay node is located anywhere between source nodes. We define the target and the minimum requirements for high-quality video communications and confirm that our proposed NC mapping rule remarkably improves the BER performance, regardless of the mobility.
This paper introduces FileTribe, a secure decentralized application for sharing files in closed groups. It employs IPFS, a distributed file system, as its data storage layer, avoiding the pitfalls of centralized storage solutions. Changes to files and group membership are recorded separately on a blockchain, making them irrefutable. A Decentralized Application (Dapp) on top of Ethereum is responsible for user authentication and reaching consensus among group members regarding the state of the group and shared files. File access control is supervised by the members themselves. These features combine to make FileTribe unique in its ability to serve a wide range of safetycritical scenarios, including ones where cloud-based or pure peerto- peer solutions fall short.
LoRa has established itself as one of the leading technologies within evolving Low Power Wide Area Networks. LoRa is primarily pillared on its patented LoRa modulation scheme that features linearly increasing chirp signals that span the LoRa bandwidth. The patented implementation of LoRa signals relies on the accurate generation of stable frequency modulated chirps using a fractional-N phase-locked loop. On the other hand, very limited research have explored alternatives for the DSP realization of LoRa-based modulation. In this paper, we focus on mathematical representations that enable the DSP generation of LoRa signals. Furthermore, we emphasize on the ability to guarantee inter-symbol phase continuity with the perspective of supporting coherent detection of LoRa signals. In conjunction with such coherent realizations, we investigate utilization of the quadrature axis with respect to each nominal LoRa signal in constructing additional orthogonal dimensions for LoRa signaling. Consequently, each transmitted extended LoRa (E-LoRa) symbol may include one extra bit when compared to nominal LoRa symbols. We build on our previous analysis for the non-coherent detection of LoRa signals in order to derive approximation formulas for BER performance of nominal LoRa under coherent detection as well as of E-LoRa. Analytical and simulation results show that coherent detection of LoRa renders approximately 0.7 dB of performance gain. On the other hand, E-LoRa supports a capacity increase that could reach up to 14% with a minimal penalty on BER performance at the scale of only 0.3 dB.
Routing in MANETs is challenging due to the time-variance of the communication channel. The reliability of the end-to-end link in such channels is improved with Opportunistic Routing (OpR) protocols, which endeavor the broadcasting nature of wireless channels using several nondisjoint propagation paths for the same traffic flow. Network coding is used together with OpR to provide multi-path coding redundancy. This combination allows to achieve optimal throughput in stationary wireless channels for mesh technologies with contention-based channel access. This paper focuses on the end-to-end throughput stability in time-variant channels. It proposes enhancements to OpR that achieve stable throughput in case of edge failures due to nodes mobility. Traces from real channel measurements are used in the evaluations. The resultant throughput is much more stable in comparison to the state-of-the-art OpR protocols.
Cloud and distributed storage applications require processing of large fragments of data. This poses memory, delay, and processing speed challenges for systems using erasure codes to reduce the cost of storage and/or increase the reliability of the system. To address these, this paper proposes and deploys designs that exploit current multi-threading capabilities of microprocessors to accelerate the encoding and decoding process of erasure codes, focusing on the case of Random Linear Network Coding (RLNC). More specifically, we propose a strategy for parallel computation based on splitting symbols into significantly smaller fragments and reordering data to speed up computation and decreasing the potential for thread blocking. We implement the strategy in C++ and carry out benchmark experiments and compare them with the single threaded state-of-the-art block RLNC encoders. We show a reduction of processing time by a factor of three by using our strategies for files of 32 MB or more as well as reducing memory usage drastically in the system. We also show that our approach provides a better scaling than the single-threaded option when increasing the number of fragments that a file is broken into. In other words, distributed storage systems can split files or data into a larger number of fragments without experiencing a speed penalty.
The interplay between Software Defined Networks (SDN) and Network Function Virtualization (NFV) provides a new networking paradigm where Virtual Network Functions (VNFs) run on physical devices. However, despite the fact that SDN and NFV enable VNFs in the optimal physical device, the optimal physical node may change over time, and the VNFs need to be moved in a seamless and reliable way. Live service migration provides the necessary tools to move virtual resources between physical hosts. The two main virtualization technologies today are virtual machines (VMs) and containers. Containers are known to boot faster than VMs, and thus, lower service downtime in the application. Despite the previous works studying the performance of the live migration process, there are still questions on how the system would behave in a real scenario where conditions are not ideal. To answer those questions, we present in this paper a testbed that performs a container live migration process under tunable conditions, namely image size, network capacity, network load, container's CPU load and container's RAM load. We evaluate both migration time and downtime for different configurations of the aforementioned parameters. We use Docker containers and a Python application in the migration process. The results showthat overloading the network can degrade the migration performance remarkably. In contrast, the impact of stressing a container inside a host is marginal.
The Age of information (AoI) was proposed in the literature to quantify the freshness of information. The majority of the work done in this area has theoretically evaluated AoI and its Peak (PAoI). In this paper, a method for obtaining the value of AoI and PAoI from experiments is proposed. We conducted an experiment emulating an M/M/1 queue and used the proposed method to evaluate AoI and PAoI. The values were compared to the expressions presented previously in the literature. Our results show that the proposed method is accurate for the M/M/1 queue. A statistical test was conducted to confirm the reliability of this conclusion.
We show that a generalization of deduplication can enable compressed storage of sensor data. The method uses error-correcting codes in a non-traditional manner to identify similar elements, and then leverages this similarity for compression. Using Reed Solomon codes, our method has a theoretical potential to reduce the cost of storing chunks of 16 bytes to as much as 5 times less, and up to 65 times less for chunks of 255 bytes. We define a simple model for sensor data, and show how our approach is able to compress data from the model, realizing its compression potential with much smaller data sets than classic deduplication requires. This demonstrates that generalized deduplication can be a viable solution for practical lossless compression of small IoT data in scenarios where classic deduplication is ineffective.
This paper analyzes the performance of cooperative caching for cellular networks by modeling the locations of base stations as a homogeneous Poisson point process, where each base station is equipped with a storage to cache the content and has cooperation with its neighboring base stations. With the proposed probabilistic cooperative caching strategy, the analytical expressions of system performance are derived and simulation results are presented to show that our analytical results closely match the numerical performance observed from the simulations. The study shows that with cooperative caching, the performance in terms of caching content delivery outage, caching content delivery delay and network congestion can indeed be improved. The paper further proposes an approximated solution for optimal probabilistic cooperative caching.
This paper describes a new design of Reed-Solomon (RS) codes when using composite extension fields. Our ultimate goal is to provide codes that remain Maximum Distance Separable (MDS), but that can be processed at higher speeds in the encoder and decoder. This is possible by using coefficients in the generator matrix that belong to smaller (and faster) finite fields of the composite extension and limiting the use of the larger (and slower) finite fields to a minimum. We provide formulae and an algorithm to generate such constructions starting from a Vandermonde RS generator matrix and show that even the simplest constructions, e.g., using only processing in two finite fields, can speed up processing by as much as two-fold compared to a Vandermonde RS and Cauchy RS while using the same decoding algorithm, and more than two-fold compared to other RS Cauchy and FFT-based RS.
Measurements of ultra-wideband (UWB) communications inside an Ariane 5 launcher are reported, assessing received signal power fluctuations and connectivity in the vehicle equipment bay. This environment is challenging due to shadowing and reflections. Our results highlight the relationship between the multipath delay spread and the UWB signal detection mechanism and indicate that UWB is a suitable candidate for radio connectivity inside launch vehicles.
This paper presents initial results regarding the potential of ultra-wideband (UWB) technology to become an enabler of self-powered industrial sensor networks. Our study uses requirements from vibration monitoring typically implemented today with wired installations. A wireless solution can significantly reduce cost if batteries do not need to be replaced often. We discuss the key factors extending network lifetime. A charge balance analysis shows that UWB is a well-suited technology for this application. Evaluations of the network lifetime at different network operating points and against alternative wireless technologies support our statement.
mmWave communication systems overcome high attenuation by using multiple antennas at both the transmitter and the receiver to perform beamforming. Upon entrance of a user equipment (UE) into a cell a scanning procedure must be performed by the base station in order to find the UE, in what is known as initial access (IA) procedure. In this paper we start from the observation that UEs are more likely to enter from some directions than from others, as they typically move along streets, while other movements are impossible due to the presence of obstacles. Moreover, users are entering with a given time statistics, for example described by inter-arrival times. In this context we propose scanning strategies for IA that take into account the entrance statistics. In particular, we propose two approaches: a memory-less random illumination (MLRI) algorithm and a statistic and memory-based illumination (SMBI) algorithm. The MLRI algorithm scans a random sector in each slot, based on the statistics of sector entrance, without memory. The SMBI algorithm instead scans sectors in a deterministic sequence selected according to the statistics of sector entrance and time of entrance, and taking into account the fact that the user has not yet been discovered (thus including memory). We assess the performance of the proposed methods in terms of average discovery time.
LoRa is a promising technology in the current Internet of Things market, which operates in un-licensed bands achieving long-range communications and with ultra power devices. In this work we capitalize on the idea introduced in [1], i.e. balance the Air-Time of the different modulation spreading factors (SF), and adapt it to operate in a typical metropolitan scenario comprising multiple gateways (GWs) interconnected to a same network server. Our proposed approach, named ADaptive Mitigation of the AIr-time pressure in lORa (AD MAIORA), relies on a suitable measure of the per-spreading-factor load at each GW - quantified by means of a so-called pressure table -, and on a relevant heuristic algorithm which attempts to balance such a per-SF-pressure. Especially in cases of very loaded scenarios, where a high number of nodes insist on the same GWs, the use of AD MAIORA shows significant performance gains, up to a factor of 5 improvements with respect to the legacy LoRaWAN's Adaptive Data Rate.
This paper presents a smart city pilot project that will be developed in the city of Bologna. Different sensor devices will be deployed in several buildings to measure environmental quantities, such as temperature, humidity, luminosity and CO2. The pilot will use the LoRa low-rate, long-range communication technology to transmit sensed data to the control center. Results of a measurement campaign to characterize the propagation in the Bologna scenario are reported. Such results are then used as input to a simulator that provides both coverage and networklevel performance of the LoRa technology in the considered scenario. The paper shows that a proper parameter setting is needed to cover large urban areas while maintaining the timeon- air sufficiently low to keep packet losses at satisfactory levels.
Availability of deep packet inspection methods and systems allows network operators to classify traffic on the base of the application type. Once classified, traffic may be subject to artificial bandwidth limitations (e.g. in case of resourcedemanding applications) or to class-dependent forwarding policies (e.g. to divert the traffic generated by specific applications on low-priority links). In this paper we describe a method that can be useful to detect the presence of class-dependent forwarding policies. The method is based on traceroute-like mechanisms embedded within the normal communication flow of an application. The method is contextualized in a study about the neutrality of mobile network operators, to understand if a correlation can be found between the presence of class-dependent forwarding strategies and limitations of bandwidth.
The Software Defined Networking paradigm (SDN) is recognized as one of the main forces that can simplify the management of wired and wireless networks. Most of the effort, so far, has been on the development of the tools and methodologies that allow network administrators and researchers to implement the desired behaviors in SDN networks. A further step in the simplification of the management is the design of a framework that can automatize the control of the network by leveraging the information gathered at the SDN controller combined with machine learning techniques. In this paper a solution is introduced that learns and predicts recurrent patterns in network load and decides which are the best policies to be implemented in order to reach a given objective. To test the proposed solution, a tool has been implemented that adapts the behavior of a wireless sensor network to reach a trade-off between energy consumption and fairness in a simulated environment that exploits the dataset produced by the SmartSantander testbed.
