Latency Constraints Research Articles

FireFace: Leveraging Internal Function Features for Configuration of Functions on Serverless Edge Platforms.

The emerging serverless computing has become a captivating paradigm for deploying cloud applications, alleviating developers' concerns about infrastructure resource management by configuring necessary parameters such as latency and memory constraints. Existing resource configuration solutions for cloud-based serverless applications can be broadly classified into modeling based on historical data or a combination of sparse measurements and interpolation/modeling. In pursuit of service response and conserving network bandwidth, platforms have progressively expanded from the traditional cloud to the edge. Compared to cloud platforms, serverless edge platforms often lead to more running overhead due to their limited resources, resulting in undesirable financial costs for developers when using the existing solutions. Meanwhile, it is extremely challenging to handle the heterogeneity of edge platforms, characterized by distinct pricing owing to their varying resource preferences. To tackle these challenges, we propose an adaptive and efficient approach called FireFace, consisting of prediction and decision modules. The prediction module extracts the internal features of all functions within the serverless application and uses this information to predict the execution time of the functions under specific configuration schemes. Based on the prediction module, the decision module analyzes the environment information and uses the Adaptive Particle Swarm Optimization algorithm and Genetic Algorithm Operator (APSO-GA) algorithm to select the most suitable configuration plan for each function, including CPU, memory, and edge platforms. In this way, it is possible to effectively minimize the financial overhead while fulfilling the Service Level Objectives (SLOs). Extensive experimental results show that our prediction model obtains optimal results under all three metrics, and the prediction error rate for real-world serverless applications is in the range of 4.25∼9.51%. Our approach can find the optimal resource configuration scheme for each application, which saves 7.2∼44.8% on average compared to other classic algorithms. Moreover, FireFace exhibits rapid adaptability, efficiently adjusting resource allocation schemes in response to dynamic environments.

Improving Worst-case TSN Communication Times of Large Sensor Data Samples by Exploiting Synchronization

Higher levels of automated driving also require a more sophisticated environmental perception. Therefore, an increasing number of sensors transmit their data samples as frame bursts to other applications for further processing. As a vehicle has to react to its environment in time, such data is subject to safety-critical latency constraints. To keep up with the resulting data rates, there is an ongoing transition to a Time-Sensitive Networking (TSN)-based communication backbone. However, the use of TSN-related industry standards does not match the automotive requirements of large timely sensor data transmission, nor it offers benefits on time-critical transmissions of single control data packets. By using the full data rate of prioritized IEEE 802.1Q Ethernet, giving time guarantees on large data samples is possible, but with strongly degraded results due to data collision. Resolving such collisions with time-aware shaping comes with significant overhead. Hence, rather than optimizing the parameters of the existing protocol, we propose a system design that synchronizes the transmission times of sensor data samples. This limits network protocol complexity and hardware requirements by avoiding tight time synchronization and time-aware shaping. We demonstrate that individual sensor data samples are transmitted without significant interference, exclusively at full Ethernet data rate. We provide a synchronous event model together with a straightforward response time analysis for synchronous multi-frame sample transmissions. The results show that worst-case latencies of such sample communication, in contrast to non-synchronized approaches, are close to their theoretical minimum as well as to simulative results while keeping the overall network utilization high.

Ultragreen Relay Transmission With Wireless Power Transfer for Advanced IoT: Session-Specific Analysis and Optimization

Reliable and energy-efficient wireless transmission is of critical importance to the success of future advanced Internet of Things (IoT). Due to the sporadic nature of IoT transmissions, the energy consumption of the individual IoT transmission session varies dramatically with the instantaneous operating environment as well as the quality of service (QoS) requirements. In this paper, we analyze and design the energyefficient relay transmission systems from an individual data transmission session perspective. Specifically, we consider a dualhop transmission system with a decode-and-forward relay that is solely powered by wireless power transfer from source node. For both time switching and power splitting modes of simultaneous power and information transmission, we analyze and minimize the total energy consumption of the system when transmitting a fixed amount of data, under a piecewise linear energy harvesting (EH) model. Closed-form expressions for optimal transmission parameters are obtained with and without the consideration of latency constraint. Through selected numerical results, we illustrate various design tradeoffs between energy consumption and latency constraint. We show that with optimal transmission parameters, relay transmission with energy transfer can achieve considerable energy saving compared to direct transmission when the direct link quality is poor and the latency constraint is not stringent. We also show that with optimized parameters, power splitting mode leads to lower energy consumption and smaller transmission duration than time switching mode, at the cost of higher implementation complexity.

MapChain-D: A Distributed Blockchain for IIoT Data Storage and Communications

With the rapid growth of the Industrial Internet of Things (IIoT) devices, managing extensive volume of IIoT data becomes a significant challenge. While the conventional cloud storage approaches with centralised data centres suffer from high latency for large-scale IIoT data storage due to the increased communications and latency overheads, distributed storage frameworks such as blockchains have become promising solutions. In this paper, we design and analyse a dual-blockchain framework for secure and scalable distributed data management in large-scale IIoT networks. The proposed framework, named <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> , consists of a data chain that is mapped to an index chain to provide efficient data storage and lookup. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> is designed for practical IIoT applications with storage, latency, and communications constraints. Detailed data exchange protocols are presented for the data insertion and retrieval operations in <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> . Based on these, theoretical analyses are provided on the space, time, and communications complexities of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> compared with conventional single-chain frameworks with local and distributed data storage. We implement our <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> prototype using open-source LoRaWAN communications with multiple RPi and Arduino devices, Kademlia-based distributed hash table (DHT), and Ethereum-based blockchain with proof-of-authority (PoA) consensus. Experimental results from our prototype show that <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> is more suitable to be deployed on resource-constrained IIoT devices. We also highlight the scalability and flexibility of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MapChain-D</i> with different numbers of edge nodes in the system.

PSTile: Perception-Sensitivity-Based 360$^\circ$ Tiled Video Streaming for Industrial Surveillance

360 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^\circ$</tex-math></inline-formula> video becomes increasingly attractive in smart factories due to its immersive experience for industrial surveillance. However, transmitting this kind of video requires a fairly large demand on the bandwidth due to its high-resolution and panoramic view. Moreover, low-latency responses of the video to users' head movements are required. This leads to the tradeoff between high video quality and low-latency response under limited bandwidth in factories. This paper proposes a perception-sensitivity (PS) based 360 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$^\circ$</tex-math></inline-formula> tiled video streaming method called PSTile for industrial surveillance. Specifically, a PS tiling strategy is constructed for valid tile grouping based on the designed PS index. Then a tile bitrate adaptation problem is formulated to allocate bitrates to valid tiles. It jointly optimizes surveillance accuracy, end-to-end latency, and quality-of-experience (QoE) of users under accuracy, latency, and bandwidth constraints. Simulations demonstrate that PSTile achieves at least 32.6% lower end-to-end latency and 19.3% higher average video bitrate than grid-tiling and clus-tiling methods.

Channel-Aware Latency Tail Taming in Industrial IoT

In this paper, we propose a novel channel-aware latency taming scheme, called Optimal Transmission Latency Taming (OTLT), to detect hidden channel state and tame the distribution tail of the packet sojourn time in Industrial Internet of Things (IIoT) devices. Specifically, we design a forward algorithm based on a hidden semi-Markov model to detect the hidden channel state, with a particular emphasis on the state sojourn duration, and to calculate the corresponding channel access probability. Then we develop a time-sensitive model to investigate the minimum sojourn time a packet spends in the IIoT device before leaving successfully. With the obtained channel access probability, the first passage probability of the proposed model is explored to find the maximum probability of a packet being successfully transmitted in a given back-off sojourn duration (BSD). The distribution tail of the packet sojourn time can be tamed by minimizing the cumulative summation of each BSD in consideration of the quadratic penalty latency constraints. Simulation results demonstrate that, in the industrial environment, the OTLT scheme can keep the packet’s sojourn duration within a quantifiable limit and variance. It can also obtain considerably efficient control over packet transmission latency in a time-varying wireless propagation channel even with the increasing number of IIoT devices.

Survey of Novel Architectures for Energy Efficient High-Performance Mobile Computing Platforms

There are many real-world applications that require high-performance mobile computing systems for onboard, real-time processing of gathered data due to latency, reliability, security, or other application constraints. Unfortunately, most existing high-performance mobile computing systems require a prohibitively high power consumption in the face of the limited power available from the batteries typically used in these applications. For high-performance mobile computing to be practical, alternative hardware designs are needed to increase the computing performance while minimizing the required power consumption. This article surveys the state-of-the-art in high-efficiency, high-performance onboard mobile computing, focusing on the latest developments. It was found that more research is needed to design high-performance mobile computing systems while minimizing the required power consumption to meet the needs of these applications.

Delay-Sensitive Service Provisioning in Software-Defined Low-Earth-Orbit Satellite Networks

With the advancement of space technology and satellite communications, low-Earth-orbit (LEO) satellite networks have experienced rapid development in the past decade. In the vision of 6G, LEO satellite networks play an important role in future 6G networks. On the other hand, a variety of applications, including many delay-sensitive applications, are continuously emerging. Due to the highly dynamic nature of LEO satellite networks, supporting time-deterministic services in such networks is challenging. However, we can provide latency guarantees for most delay-sensitive applications through data plane traffic shaping and control plane routing optimization. This paper addresses the routing optimization problem for time-sensitive (TS) flows in software-defined low-Earth-orbit (LEO) satellite networks. We model the problem as an integer linear programming (ILP) model aiming to minimize path handovers and maximum link utilization while meeting TS flow latency constraints. Since this problem is NP-hard, we design an efficient longest continuous path (LCP) approximation algorithm. LCP selects the longest valid path in each topology snapshot that satisfies delay constraints. An auxiliary graph then determines the routing sequence with minimized handovers. We implement an LEO satellite network testbed with Open vSwitch (OVS) and an open-network operating system (ONOS) controller to evaluate LCP. The results show that LCP reduces the number of path handovers by up to 31.7% and keeps the maximum link utilization lowest for more than 75% of the time compared to benchmark algorithms. In summary, LCP achieves excellent path handover optimization and load balancing performance under TS flow latency constraints.

Efficient Distributed DNNs in the Mobile-Edge-Cloud Continuum

In the mobile-edge-cloud continuum, a plethora of heterogeneous data sources and computation-capable nodes are available. Such nodes can cooperate to perform a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">distributed</i> learning task, aided by a learning controller (often located at the network edge). The controller is required to make decisions concerning (i) data selection, i.e., which data sources to use; (ii) model selection, i.e., which machine learning model to adopt, and (iii) matching between the layers of the model and the available physical nodes. All these decisions influence each other, to a significant extent and often in counter-intuitive ways. In this paper, we formulate a problem addressing all of the above aspects and present a solution concept called RightTrain, aiming at making the aforementioned decisions in a joint manner, minimizing energy consumption subject to learning quality and latency constraints. RightTrain leverages an expanded-graph representation of the system and a delay-aware Steiner tree to obtain a provably near-optimal solution while keeping the time complexity low. Specifically, it runs in polynomial time and its decisions exhibit a competitive ratio of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2(1+\epsilon)$</tex-math> </inline-formula> , outperforming state-of-the-art solutions by over 50%. Our approach is also validated through a real-world implementation.

Reliable Dynamic Service Chain Scheduling in 5G Networks

As a key enabler of future 5G network, Service Function Chain (SFC) forwards the traffic flow along a chain of Virtual Network Functions (VNFs) to provide network services flexibility. One of the most important problems in SFC is to deploy the VNFs and schedule arriving requests among computing nodes to achieve low latency and high reliability. Existing works consider a static network and assume that all SFC requests are known in advance, which is impractical. In this paper, we focus on the dynamic 5G network environment where the SFC requests arrive randomly and the computing nodes can redeploy all types of VNF with a time cost. We formulate the problem of SFC scheduling in NFV-enabled 5G network as a mixed integer non-linear programing. The objective is to maximize the number of requests satisfying the latency and reliability constraints. To solve the problem, we propose an efficient algorithm to decide the redundancy of the VNFs while minimizing delay. Then we present a state-of-art Reinforcement Learning (RL) to learn SFC scheduling policy to increase the success rate of SFC requests. The evaluation result shows that our proposed RL solution can increase the success rate by 18.7% over the benchmark algorithms.

RASM: Resource-Aware Service Migration in Edge Computing based on Deep Reinforcement Learning

Multi-access Edge Computing (MEC) paradigm allows devices to offload their intensive service tasks that require high Quality of Experience (QoE). Devices mobility forces services to migrate between MECs to maintain QoE in terms of delay. The decision on when to migrate a service requires a cost and QoE tradeoff, and destination MEC selection needs to be done upon latency and resource availability constraints to minimize migrations. To this end, we propose a novel Resource-Aware Service Migration (RASM) mechanism using Deep Q-Network (DQN) to make migration decisions by achieving tradeoff between the QoE in terms of delay and migration cost. Moreover, DQN learns the best policy for maximizing QoE by selecting the migration destination based on the MECs proximity to the device and estimated resource availability at the servers using queuing model. Results show faster convergence to optimal policy, reduced average end-to-end service delay by 27%, and smaller service rejection rate by 24% comparing to the state-of-the-art.

Intelligent Resource Allocation for V2V Communication with Spectrum-Energy Efficiency Maximization.

Aiming to address the limitations of traditional resource allocation algorithms in the Internet of Vehicles (IoV), whereby they cannot meet the stringent demands for ultra-low latency and high reliability in vehicle-to-vehicle (V2V) communication, this paper proposes a wireless resource allocation algorithm for V2V communication based on the multi-agent deep Q-network (MDQN). The system model utilizes 5G network slicing technology as its fundamental feature and maximizes the weighted spectrum-energy efficiency (SEE) while satisfying reliability and latency constraints. In this approach, each V2V link is treated as an agent, and the state space, action, and reward function of MDQN are specifically designed. Through centralized training, the neural network parameters of MDQN are determined, and the optimal resource allocation strategy is achieved through distributed execution. Simulation results demonstrate the effectiveness of the proposed scheme in significantly improving the SEE of the network while maintaining a certain success rate for V2V link load transmission.

Joint task offloading and resource allocation for multi-user and multi-server MEC networks: A deep reinforcement learning approach with multi-branch architecture

Mobile Edge Computing (MEC) is a promising computing paradigm in the context of 5G networks, as it enables the migration of workloads from User Equipments (UEs) to nearby MEC servers, thereby providing additional computing resources to UEs. In this paper, we propose a joint optimization approach to offloading decisions and resource allocation in a multi-user and multi-server MEC system, which operates in a time-varying environment. Our objective is to minimize the average task latency and discard rate under the constraints of latency and limited computing resources of the server. While traditional optimization methods have been used to solve computational offloading problems in static environments, these methods are not suitable for time-varying systems. Deep reinforcement learning can be an effective method for solving optimization problems in time-varying environments, as it can be used to adjust strategies in real-time in response to changes in the environment. However, the increased number of UEs in the system leads to a combinatorial increase in the number of possible actions, making it difficult for the algorithm to learn. To address this issue, we propose a multi-branch network based Deep Q Network (DQN) algorithm called Branch Deep Q Network (BDQN), which modifies the action generation network into a multi-branch network structure, each branch generates one-dimensional action. This modification makes the number of network outputs increase linearly. Numerical results show that the BDQN algorithm outperforms other baseline algorithms in terms of average task latency and discard rate.

A joint cluster-based RRM and Low-latency framework using the full-duplex mechanism for NR-V2X networks

The radio resource management (RRM) problem in new radio vehicle-to-everything (NR-V2X) communication systems is a combinatorial optimization problem, and it is hard to achieve an optimum result in polynomial time. To reduce the complexity, linear algorithms or meta-heuristics can solve the problem. In this study, we proposed a joint cluster-based resource management and low-latency framework using a full-duplex mechanism (JCRRM-FD) for NR-V2X networks, which is used for vehicle-to-vehicle (V2V)/ device-to-device (D2D) and vehicle-to-infrastructure (V2I) transmission in NR-V2X networks. The delay, resource management, and system throughput analysis of the full-duplex mechanism in this framework was performed to deal with resource utilization and latency constraints. Moreover, joint cluster-based radio resource management and ant colony optimization (ACO) algorithms were proposed to manage and optimize resources efficiently to achieve user separation. The swarm intelligence algorithm is a standard meta-heuristic algorithm employed to deal with the optimization concern by exploiting the V2X communication network’s unlimited speed (maximizing the data transfer capacity and overall network performance) while considering the quality-of-service (QoS) requirements. A comprehensive experiment analysis was enacted to evaluate the efficiency of the developed JCRRM-FD framework with baseline approaches. Based on the simulation results, the proposed JCRRM-FD framework enhances the fairness index, average delay, packet drop rate, best cost value, CDF, and throughput compared to the benchmark approaches.

Fortified-Chain 2.0: Intelligent Blockchain for Decentralized Smart Healthcare System

The Internet of Medical Things (IoMT) technology’s fast advancements aided smart healthcare systems to a larger extent. IoMT devices, on the other hand, rely on centralized processing and storage systems because of their limited computational and storage capacity. The reliance is susceptible to a single point of failure (SPoF) and erodes the user control over their medical data. In addition, Cloud models result in communication delays, which slow down the system’s overall reaction time. To overcome these issues a decentralized distributed smart healthcare system is proposed that eliminates the SPoF and third-party control over healthcare data. Additionally, the proposed Fortified-Chain 2.0 uses a blockchain-based selective sharing mechanism with a mutual authentication technique to solve the issues such as data privacy, security and trust management in decentralized peer-to-peer healthcare systems. Also, we suggested a hybrid computing paradigm to deal with latency, computational and storage constraints. A novel distributed Machine Learning (ML) module named Random Forest Support Vector Machine (RFSVM) also embedded into the Fortified-Chain 2.0 system to automate patient health monitoring. In the RFSVM module, Random Forest (RF) is used to select an optimal set of features from patients data in real time environment and also Support Vector Machine (SVM) is used to perform the decision making tasks. The proposed Fortified-Chain 2.0 works on a private blockchain-based Distributed Data Storage System (DDSS) that improves the system-level transparency, integrity, and traceability. Fortified-Chain 2.0 outperformed the existing Fortified-Chain in terms of low latency, high throughput, and availability with the help of a mutual authentication method.

Hybrid Interval Type-2 Fuzzy AHP and COPRAS-G-based trusted neighbour node Discovery in Wireless Sensor Networks

In Wireless Sensor Networks (WSNs), reliable and rapid neighbour node discovery is considered as the crucial operation which frequently needs to be executed over the entire lifecycle. Several neighbour node discovery mechanisms are proposed for reducing the latency or extending the sensor nodes’ lifetime. But majority of the existing neighbour node discovery mechanisms failed in addressing the critical issues of real WSNs related to energy consumptions, constraints of latency, uncertainty of node behaviors, and communication collisions. In this paper, Hybrid Interval Type-2 Fuzzy Analytical Hierarchical Process (AHP) and Complex Proportional Assessment using Grey Theory (COPRAS-G)-based trusted neighbour node discovery scheme (FAHPCG) is proposed for better data dissemination process. In specific, Interval Type 2 Fuzzy AHP is applied for determining the weight of the evaluation criteria considered for neighbour node discovery, and then Grey COPRAS method is adopted for prioritizing the sensor nodes of the routing path established between the source and destination. It adopted the merits of fuzzy theory for handling the uncertainty and vagueness involved in the change in the behavior of sensor nodes during the process of neighbour discovery. It is proposed with the capability of exploring maximized number of factors that aids in exploring the possible dimensions of sensor nodes packet forwarding potential during the process of neighbour node discovery. The simulation results of the proposed FAHPCG scheme confirmed an improved neighbour node discovery rate of 23.18% and prolonged the sensor nodes lifetime to the maximum of 7.12 times better than the baseline approaches used for investigation.

Enabling Real-Time Remote Monitoring of Ships by Lossless Protocol Transformations

This paper uses data processing techniques to reduce the required transmission bandwidth in ship-to-shore communications. The proposed framework (ONline Efficient Sources Transmission Optimizer -) leverages state-of-the-art technologies and novel algorithms to automatically optimize transmissions under structural (e.g., available bandwidth, fixed packet overhead) and user-defined (e.g., maximum latency) constraints. In addition, authenticates and encrypts the communication between the ship and the shore via mainstream free and open-source software components. Initially, we present the abstract mathematical formulation of the problem, with its assumptions, goal function, constraints, and significant quantities. Then, we introduce the architecture of a system capable of continuously estimating the compressibility, processing and transmission time of streaming data. Such estimations allow to calculate and apply optimal parameters for achieving the best compression ratio. Lastly, using a prototypical implementation, we evaluate the system performance with a Class B ship simulator on two realistic use cases. Our experiments show an excellent compression ratio with maritime protocols (more than 40:1) and a limited latency impact, demonstrating the approach’s viability.

Joint Computation Offloading and Resource Allocation for MIMO-NOMA Assisted Multi-User MEC Systems

This paper investigates the resource allocation and computation offloading problem for multi-access edge computing (MEC) systems, where multiple mobile users (MUs) equipped with multiple antennas access the base station in a non-orthogonal multiple access manner. We jointly optimize the offloading ratio, computational frequency and transmit precoding matrix of each MU to minimize the total energy consumption of all MUs while satisfying the latency constraints. The problem is formulated as a non-convex optimization problem and a two-layer iterative method is proposed to solve the problem efficiently with low complexity. Specifically, we first decompose the original problem into several subproblems, and then sequentially solve these subproblems in an alternative fashion. Furthermore, we also discuss the optimal decoding order of MUs under two different scenarios. Firstly, when the MUs’ channel conditions are similar, by deriving closed-form expressions for energy consumptions of all MUs, we prove that the optimal decoding order is only determined by the latency requirements. On the other hand, when the MUs’ channel conditions are different, we show that the optimal decoding order is determined by both the channel conditions and the latency requirements. As such, we propose a metric aiming to balance the effects of channel conditions and latency requirements on the MUs’ decoding order. Simulation results validate the convergence of the proposed method and demonstrate its superiority over benchmark algorithms.

Fronthaul latency and capacity constrained cost-effective and energy-efficient 5G C-RAN deployment

The third-generation partnership project (3GPP) has identified different new functional split points, defined as options 1–7, between the baseband unit (BBU) and the remote unit (RU) for the 5G centralized radio access networks (C-RANs). The functional splits facilitate lower fronthaul line rate and latency requirements while imposing technical complexity and cost-efficiency trade-offs. Among all the functional split options, the standardization groups such as 3GPP, Open RAN (O-RAN), and small cell forum have selected option-7 as the most suitable choice for the 5G C-RAN deployment that can meet the operator’s requirements. One of the biggest challenges for designing option-7 functional split-based C-RAN is the capacity and latency requirement fulfillment as specified by 3GPP, while minimizing the deployment cost. In this regard, we propose an optimization model for cost-effective and energy-efficient option-7 functional split-based C-RAN architecture using wavelength division multiplexed passive optical network (WDM-PON) in the fronthaul. For the given locations of the next-generation core (NGC) and the set of RUs, the proposed optimization model explores the most suitable locations for the placement of BBU pools and arrayed waveguide gratings (AWGs). It provides complete interconnection details among NGC, BBU pools, AWGs, and RUs. It also provides the optimal number of digital units in each selected BBU pool, the splitting ratio of each AWG, and maximum fronthaul reach, satisfying the fronthaul latency, capacity, and power budget constraints. To the best of our knowledge, for the first time, we estimate the worst-case queuing delay for the upstream traffic flow using the G/G/1 queuing model and investigate its impact on fronthaul latency and deployment cost for different RU configurations. We compare the performance of the proposed option-7 functional split-based C-RAN with the traditionally used option-8 functional split-based C-RAN in terms of the deployment cost and fronthaul latency improvement.

Not All Neighbors Matter: Point Distribution-Aware Pruning for 3D Point Cloud

Applying deep neural networks to 3D point cloud processing has demonstrated a rapid pace of advancement in those domains where 3D geometry information can greatly boost task performance, such as AR/VR, robotics, and autonomous driving. However, as the size of both the neural network model and 3D point cloud continues to scale, reducing the entailed computation and memory access overhead is a primary challenge to meet strict latency and energy constraints of practical applications. This paper proposes a new weight pruning technique for 3D point cloud based on spatial point distribution. We identify that particular groups of neighborhood voxels in 3D point cloud contribute more frequently to actual output features than others. Based on this observation, we propose to selectively prune less contributing groups of neighborhood voxels first to reduce the computation overhead while minimizing the impact on model accuracy. We apply our proposal to three representative sparse 3D convolution libraries. Our proposal reduces the inference latency by 1.60× on average and energy consumption by 1.74× on NVIDIA GV100 GPU with no loss in accuracy metric