Annals of Telecommunications最新文献

Specific workloads are increasingly offloaded to accelerators such as a graphic processing unit (GPU) and field-programmable gate array (FPGA) for real-time processing and computing efficiency. Because accelerators are expensive and consume much power, it is desirable to increase the efficiency of accelerator utilization by sharing accelerators among multiple servers over a network. However, task offloading over a network has the problem of latency due to network processing overhead in remote offloading. This paper proposes a low-latency system for accelerator offloading over a network. To reduce the overhead of remote offloading, we propose a system composed of (1) fast recombination processing of chunked data with a simple protocol to reduce the number of memory copies, (2) polling-based packet receiving check to reduce overhead due to interrupts in interaction with a network interface card, and (3) a run-to-completion model in network processing and accelerator offloading to reduce overhead with context switching. We show that the system can improve performance by 66.40% compared with a simple implementation using kernel protocol stack and confirmed the performance improvement with a virtual radio access network use case as a low-latency application. Furthermore, we show that this performance can also be achieved in practical usage in data center networks.

The interplay of physical layer enhancements and classic random access protocols is the objective of this paper. Successive interference cancellation (SIC) is among the major enhancements of the physical layer. Considering the classic representatives of random access protocols, Slotted ALOHA and Channel Sensing Multiple Access (CSMA), we show that two regimes can be identified as a function of the communication link spectral efficiency. In case of high levels of spectral efficiency, multi-packet reception enabled by SIC is of limited benefit. Sum-rate performance is dominated by the effectiveness of the Medium Access Control (MAC) protocol. On the contrary, for low spectral efficiency levels, sum-rate performance is essentially dependent on physical layer SIC capability, while the MAC protocol has a marginal impact. Limitations due to transmission power dynamic range are shown to induce unfairness among nodes. However, the unfairness issue fades away when the system is driven to work around the sum-rate peak achieved for low spectral efficiency. This can also be confirmed by looking at Age of Information (AoI) metric. The major finding of this work is that SIC can boost performance, while still maintaining a fair sharing of the communication channel among nodes. In this regime, the MAC protocol appears to play a marginal role, while multi-packet reception endowed by SIC is prominent to provide high sum-rate, low energy consumption, and low AoI.

We present Soteria, a data processing pipeline for detecting multi-institution attacks. Soteria uses a set of machine learning techniques to detect future attacks, predict their future targets, and rank attacks based on their predicted severity. Our evaluation with real data from Canada-wide academic institution networks shows that Soteria can predict future attacks with 95% recall rate, predict the next targets of an attack with 97% recall rate, and detect attacks in the first 20% of their life span. Soteria is deployed in production and is in use by tens of Canadian academic institutions that are part of the CANARIE IDS project.

The increasing development of cryptocurrencies has brought cryptojacking as a new security threat in which attackers steal computing resources for cryptomining. The digitization of the supply chain is a potential major target for cryptojacking due to the large number of different infrastructures involved. These different infrastructures provide information sources that can be useful to detect cryptojacking, but with a wide variety of data formats and encodings. This paper describes the semantic data aggregator (SDA), a normalization and aggregation system based on data modelling and low-latency processing of data streams that facilitates the integration of heterogeneous information sources. As a use case, the paper describes a cryptomining detection system (CDS) based on network traffic flows processed by a machine learning engine. The results show how the SDA is leveraged in this use case to obtain aggregated information that improves the performance of the CDS.

Non-orthogonal multiple access (NOMA) techniques have the potential to achieve large connectivity requirements for future-generation wireless communication. NOMA detection techniques require conventional successive interference cancellation (SIC) techniques for uplink and downlink transmissions on the receiver side to decode the transmitted signals. Multipath fading significantly impacts the SIC process and correct signal detection due to propagation delay and fading channel. Deep learning (DL) techniques can overcome conventional SIC detection limitations. Signal detection for a multi-user NOMA wireless communication system that relies on orthogonal frequency-division multiplexing (OFDM) is discussed using various DL approaches in this paper. For multi-user signal detection, different deep learning-based sequential model neural networks, gated recurrent unit (GRU), long short-term memory (LSTM), and bi-directional long short-term memory (Bi-LSTM) are applied. The deep neural network is initially trained offline with multi-user NOMA signals in the OFDM system and used to recover transmitted signals directly. DL-based sequential models with different cyclic prefixes and fast Fourier transforms with various M-phase shift keying (M-PSK) modulation schemes are discussed with deep learning optimization algorithms. In simulation results, the conventional SIC technique with minimum mean square error approach is compared to the effectiveness of DL-based models for signal detection of multi-user NOMA systems by their bit error rate performances. The root mean square error performance of different deep learning-based sequence models with other optimizers is also discussed. Moreover, the robustness of the Bi-LSTM is evaluated with the reliability of other DL-based sequential model applications in the multi-user downlink NOMA wireless communication systems.

While GPS has traditionally been the primary positioning technology, 3GPP has more recently begun to include positioning services as native, built-in features of future-generation cellular networks. With Release 16 of the 3GPP, finalized in 2021, a significant standardization effort has taken place for positioning in 5G networks, especially in terms of physical layer signals, measurements, schemes, and architecture to meet the requirements of a wide range of regulatory, commercial and industrial use cases. However, experimentally driven research aiming to assess the real-world performance of 5G positioning is still lagging behind, root causes being (i) the slow integration of positioning technologies in open-source 5G frameworks, (ii) the complexity in setting up and properly configuring a 5G positioning testbed, and (iii) the cost of a multi-BS deployment. This paper sheds some light on all such aspects. After a brief overview of state of the art in 5G positioning and its support in open-source platforms based on software-defined radios (SDRs), we provide advice on how to set up positioning testbeds, and we demonstrate, via a set of real-world measurements, how to assess aspects such as reference signal configurations, localization algorithms, and network deployments. Our contribution further includes an assessment of the efficacy of utilizing measurements obtained from a single-link limited-size testbed to forecast localization performance in more elaborate (and hence more expensive) multi-node network settings. We posit that our methodological insights can assist in lowering the entry cost barriers associated with conducting 5G positioning experiments and, consequently, promote additional experimental research in this domain.

In the wireless channel state information (CSI)-assisted amplify-and-forward (AF) networks, an instantaneous CSI of the first hop is required to scale the amplification gain. However, the deployment of instantaneous CSI always remains difficult in real applications because it increases the CSI overhead and causes resources wastage such as power and bandwidth. In order to reduce the CSI overhead and the system complexity, we suggest the integration of semi-blind relay in secure non-orthogonal multiple access (NOMA) systems where only a statistical CSI of the first hop is used to generate the amplification gain. This paper addresses the performance analysis of the secure semi-blind AF-NOMA (S-SBAF-NOMA) schemes in which the base station communicates with a pair of users via a semi-blind relay node in the presence of one eavesdropper. First, we provide the expressions for the end-to-end signal-to-noise ratio (SNR) at each receiver node. We then derive new analytical and asymptotic expressions for strictly positive secrecy capacity (SPSC) and secrecy outage probability (SOP). To ensure the exactness and the tractability of mathematical analysis, we provide some numerical results obtained through simulation rounds in Matlab, and we compare them with those of secure CSI-assisted AF-NOMA (S-CSIAF-NOMA) networks. Our results show that the proposed S-SBAF-NOMA scheme achieves comparable secrecy performance/same performance bounds as compared to S-CSIAF-NOMA scheme at the gain of a decrease in processing complexity and system overhead. Numerical results also demonstrate that S-SBAF-NOMA networks achieve superior secrecy performance for lower values of target data rates and SNR of the illegal link.

This paper describes the effects of running in-network quality adaption by trimming the packets of layered video streams at the edge. The video stream is transmitted using the BPP transport protocol, which is like UDP, but has been designed to be both amenable to trimming and to provide low-latency and high reliability. The traffic adaption uses the Packet Wash process of Big Packet Protocol (BPP) on the transmitted Scalable Video Coding (SVC) video streams as they pass through a network function which is BPP-aware and embedded at the edge. Our previous work has either demonstrated the use of Software Defined Networking (SDN) controllers to implement Packet Wash directly, or the use of a network function in the core of the network to do the same task. This paper presents our effort to deploy and evaluate such a process at the edge, highlighting the packet trimming algorithm and showing the packet trimming effects on the streams. We compare the performance of transmitting video using BPP and the Packet Wash trimming, against alternative transmission schemes, namely UDP and HTTP adaptive streaming (HAS), presenting a number of quality parameters. The results demonstrate that providing traffic engineering using in-network quality adaption using packet trimming, provides high quality at the receiver.

Wireless sensor networks (WSNs) sense and collect information from a desired phenomenon with the help of sensor nodes that have limited computational power, battery, and memory. Several data aggregation approaches are proposed to make the sensor networks energy-efficient, increasing the network’s lifetime by controlling data redundancy at aggregator nodes. Redundant data is suppressed before transmission to the sink. In this work, our aim is to enhance the network lifetime by efficiently utilizing the network’s energy through controlled data redundancy and minimizing data transmission to the sink. Data aggregation occurs in two steps: firstly, within clusters where the cluster-head serves as the aggregation point, and secondly, at a central point in the network where the gateway node acts as the aggregation point. Experiments demonstrate that our proposed approach yields better results compared to a benchmark clustering protocol in terms of network stability, the number of data packets transferred to the destination, energy dissipation of nodes, and overall network lifetime.

Named-Data Networking (NDN) over Low Power and Lossy Networks (LLNs), employing IEEE 802.15.4 communication technology, is projected to provide native support for mobility and efficient content delivery for the emerging Internet of Things (IoT). While many interest forwarding strategies have been proposed for NDNs over LLNs, most existing studies have relied on software simulations to evaluate their performance due to the lack of analytical modeling tools. This paper introduces the first analytical model for estimating the Interest Satisfaction Ratio (ISR) in NDN over LLNs, which is a crucial metric for assessing the effectiveness of interest forwarding strategies. We develop the analytical model specifically for the broadcast forwarding strategy, which has been extensively studied due to its simplicity and ease of implementation. Simulation results confirm that the proposed model predicts the ISR with reasonable accuracy. The model is then used to elucidate the strong interaction between the CSMA/CA parameters of the IEEE 802.15.4 standard and the achieved ISR.