awesome-edge-computing
A curated list of awesome edge computing, including Frameworks, Simulators, Tools, etc.
https://github.com/qijianpeng/awesome-edge-computing
Last synced: 1 day ago
JSON representation
Edge AI Hardware Products
Engine
- Apache OpenWhisk - Apache OpenWhisk is an opensource, distributed Serverless platform that executes functions (fx) inresponse to events at any scale. OpenWhisk manages the infrastructure, serversand scaling using Docker containers so you can focus on building amazing andefficient applications.
- Bochs - Bochs is a highly portable open sourceIA-32 (x86) PC emulator written in C++, that runs on most popular platforms.It includes emulation of the Intel x86 CPU, common I/O devices, and a customBIOS. Bochs can be compiled to emulate many different x86 CPUs, from early 386to the most recent x86-64 Intel and AMD processors which may even not reachedthe market yet.
- BOINC - BOINC lets you help cutting-edge scienceresearch using your computer. The BOINC app, running on your computer,downloads scientific computing jobs and runs them invisibly in the background.
- Chameleon - Chameleon is a large-scale,deeply reconfigurable experimental platform built to support Computer Sciencessystems research. Community projects range from systems research developingnew operating systems, virtualization methods, performance variabilitystudies, and power management research to projects in software definednetworking, artificial intelligence, and resource management.
- EdgeX Foundry - EdgeX Foundry is a highlyflexible and scalable open source software framework that facilitatesinteroperability between devices and applications at the IoT Edge.
- Folding@home - Folding@home (FAH or F@h) is adistributed computing project for simulating protein dynamics, including theprocess of protein folding and the movements of proteins implicated in avariety of diseases. It brings together citizen scientists who volunteer torun simulations of protein dynamics on their personal computers and scientificresearchers. Insights from these data are helping scientists to betterunderstand biology and providing new opportunities for developingtherapeutics.
- K3s - Lightweight Kubernetes. Production ready, easy toinstall, half the memory, all in a binary less than 100 MB. Great for Edge,IoT, CI, Development, ARM, Embedding k8s, and Situations where a PhD in k8sclusterology is infeasible
- Krustlet - Krustlet: Kubernetes Kubelet in Rust forrunning WASM Krustlet acts as a Kubelet by listening on the event stream fornew pods that the scheduler assigns to it based on specific Kubernetestolerations. The default implementation of Krustlet listens for thearchitecture wasm32-wasi and schedules those workloads to run in awasmtime-based runtime instead of a container runtime.
- wasmCloud - wasmCloud is a distributed platform forwriting portable business logic that can run anywhere from the edge to thecloud. Secure by default, wasmCloud aims to strip wasteful boilerplate fromthe developer experience and return joy to the act of building distributedapplications.
- WasmEdge Runtime - WasmEdge (previously known as SSVM)is a lightweight, high-performance, and extensible WebAssembly runtime forcloud native, edge, and decentralized applications. It is the fastest Wasm VMtoday. WasmEdge is an official sandbox project hosted by the CNCF. Its usecases include serverless apps, embedded functions, microservices, smartcontracts, and IoT devices.
- Wasmer - Wasmer is a fast and secure WebAssembly runtimethat enables super lightweight containers to run anywhere: from Desktop to theCloud, Edge and IoT devices.
- Wasmtime - Wasmtime is a Bytecode Alliance projectthat is a standalone wasm-only optimizing runtime for WebAssembly and WASI. Itruns WebAssembly code outside of the Web, and can be used both as acommand-line utility or as a library embedded in a larger application.
- AREG SDK - AREG SDK is aninterface-centric real-time asynchronous communication engine to enabledistributed- andmist-computing,where connected Things interact and provide services as if they act like thindistributed servers. The communication engine of AREG SDK is based on ObjectRemote Procedure Call (ORPC) protocol.
- Baetyl - Baetyl is an open edge computingframework of Linux Foundation Edge that extends cloud computing, data andservice seamlessly to edge devices. It can provide temporary offline,low-latency computing services include device connection, message routing,remote synchronization, function computing, video capture, AI inference,status reporting, configuration ota etc.
- ENORM - The ENORM framework primarilyaddresses the deployment and load balancing challenges on individual edgenodes. ENORM operates in a three-tier environment, but a master controllerdoes not control the edge nodes. Instead, it is assumed that they are visibleto cloud servers that may want to make use of the edge. The framework allowsfor partitioning a cloud server and offloading it to edge nodes for improvingthe overall QoS of the application.
- EdgeMesh - :EdgeMesh is a part ofKubeEdge, and provides a simple network solution for the inter-communicationsbetween services at edge scenarios.
- FogFlow - FogFlow is an IoT edgecomputing framework to automatically orchestrate dynamic data processing flowsover cloud and edges driven by context, including system context on theavailable system resources from all layers, data context on the registeredmetadata of all available data entities, and also usage context on theexpected QoS defined by users.
- KubeEdge - :KubeEdge is an open sourcesystem extending native containerized application orchestration and devicemanagement to hosts at the Edge. It is built upon Kubernetes and provides coreinfrastructure support for networking, application deployment and metadatasynchronization between cloud and edge. It also supports MQTT and allowsdevelopers to author custom logic and enable resource constrained devicecommunication at the Edge. KubeEdge consists of a cloud part and an edge part.
-
OCI
- a framework that enables networkoperators with the ability to open up their edge facilities to ApplicationService Providers, by offering edge computing. As an example, consider athird-party Application Service Provider selling an IoT device (say, a homethermostat or security camera). Whenever that device shows up at the edge ofthe network, the Open Carrier Interface framework starts the ApplicationService Provider implemented edge software at the network operator’s CentralOffices, which supports the IoT device with edge processing. Thus, even anearly-stage Application Service Provider, who has few financial or physicalresources, can offer the same level of edge support as giant companies, andtherefore can compete with them on an even footing. read
more. - OpenYurt - OpenYurt has been designedto meet various DevOps requirements against typical edge infrastructures. Itprovides the same user experience for managing the edge applications as ifthey were running in the cloud infrastructure. It addresses specificchallenges for cloud-edge orchestration in Kubernetes such as unreliable ordisconnected cloud-edge networking, edge node autonomy, edge devicemanagement, region-aware deployment and so on. OpenYurt preserves intactKubernetes API compatibility, is vendor agnostic, and more importantly, isSIMPLE to use.
- PhoneSploit Pro - Anall-in-one hacking tool to remotely exploit Android devices using ADB andMetasploit-Framework to get a Meterpreter session.
- SuperEdge - SuperEdge is an opensource container management system for edge computing to manage computeresources and container applications in multiple edge regions. These resourcesand applications, in the current approach, are managed as one singleKubernetes cluster. A native Kubernetes cluster can be easily converted to aSuperEdge cluster.
-
Distributed Strom
- DistributedStorm is an extension of Apache Storm that supports the execution ofdistributed, adaptive, and QoS-aware scheduling algorithms. It introduces somekey components on each worker node that enhance the system with adaptationcapabilities, relying on a MAPE (Monitor, Analyze, Plan, and Execute) feedbackloop periodically executed. Specifically, the newly introduced components are:the AdaptiveSchedulers, the QoSMonitors, and the WorkerMonitors. Logics of a
Boltcan be regarded as an edge server processes the data. - WSO2-IoT Server - An extension of the popularopen-source enterprise service-oriented integration platform WSO2 server thatconsists of certain IoT-related mechanisms, such as connecting a broad rangeof common IoT devices with the cloud using standard protocols such as MQTT andXMPP. Further, WSO2–IoT server includes the embedded Siddhi 3.0 component thatallows the system to deploy real-time streaming processes in embedded devices.In other words, WSO2–IoT server provides the FEC computing capability toouter-edge devices.
-
AimDB
- Data ingestion layer fordistributed systems with typed contracts, safe schema evolution and one placeto see and manage every node, from microcontroller to cloud. Runs on
no_stdmicrocontrollers (Embassy) and Tokio. - KubeStellar Console - Open source AI-poweredmulti-cluster Kubernetes dashboard for hybrid edge and cloud environments. Providesreal-time observability across edge and cloud clusters with 20+ CNCF integrations(Argo, Kyverno, Prometheus, Grafana, Istio, Flux, Falco, OPA/Gatekeeper), AI-guidedcluster operations, and unified management of distributed edge workloads. CNCF Sandboxproject (Apache 2.0).
- FogAtlas - FogAtlas (evolution of the former Foggyplatform) is a software framework aiming to manage a geographicallydistributed and decentralized Cloud Computing infrastructure that providescomputational, storage and network services close to the data sources and theusers, embracing the Fog Computing paradigm. FogAtlas is able to manage the socalled Cloud-to-Thing Continuum offering service-aware workload placement andzero-touch deployment. It is an evolution of the well known paradigms of IaaSand PaaS adding the concept of “locality” to the traditional Cloud Computingmodel and easing the operations of a Fog Computing infrastructure.
- ForestHub edge-agents - The ~30MB open-source edge AI agent runtime from ForestHub.Runs offline on Linux edge devices (Raspberry Pi, Jetson, industrialgateways), with GPIO/UART/MQTT as first-class nodes and local SLMs alongsidecloud LLMs. Engine ships as a distroless container; visual builder included.
- Xybrid - An open-source, local-firstruntime for running LLMs, ASR, and TTS natively in apps and games.
- Acurast - A decentralized compute network built onsmartphones. Developers deploy Node.js or native Linux workloads to phonesaround the world, and the phones' Trusted Execution Environments make theexecution verifiable and confidential.
- EdgeGallery (Documents maintained inChinese) - EdgeGallery is a 5G edge computingopen source project supported by device vendors, operators, vertical industrypartners, etc. Its goal is to create a public platform for edge computing thatmeets the characteristics of 5G MEC "connectivity and computation",standardize open network capabilities (especially for 5G networks), andgeneralize the lifecycle-management process of MEC application development(testing, migration, and operation).
- Spinifex - An open-source,AWS-compatible cloud for bare-metal, edge and on-premises deployments. Itruns EC2, EBS, S3, VPC and IAM on KVM with no external control plane, keepsrunning through network disconnection, and works with the existing AWS CLI,SDKs and Terraform. Docs
- WebAssembly: Curated list of awesome things regarding WebAssembly (wasm)ecosystem.
- WebAssembly MicroRuntime - WebAssemblyMicro Runtime (WAMR) is a standalone WebAssembly (WASM) runtime with a smallfootprint. It includes a few parts as below: 1, The "iwasm" VM core,supporting WebAssembly interpreter, ahead of time compilation (AoT) andJust-in-Time compilation (JIT). 2, The application framework and thesupporting API's for the WASM applications. 3, The dynamic management of theWASM applications
Networks
- Awesome-pcaptools - :Acollection of tools developed by other researchers in the Computer Sciencearea to process network traces.
- EdgeNet - edgeNet is a distributed edgecloud, in the family of PlanetLab, GENI, Canada’s SAVI infrastructure, Japan’sJGN-X, Germany’s G-Lab, and PlanetLab Europe. It is a modern distributed edgecloud, incorporating advances in Cloud technologies over the past few years.
- Naming Data Network Platform - NDNis an entirely new architecture, but one whose design principles are derivedfrom the successes of today’s Internet, reflecting our understanding of thestrengths and limitations of the current Internet architecture, and one thatcan be rolled out through incremental deployment over the current operationalInternet.
- Node-RED - Node-RED is a programming tool for wiringtogether hardware devices, APIs and online services in new and interestingways. It provides a browser-based editor that makes it easy to wire togetherflows using the wide range of nodes in the palette that can be deployed to itsruntime in a single-click.
- Open vSwitch - is a production quality,multilayer virtual switch licensed under the open source Apache 2.0 license.It is designed to enable massive network automation through programmaticextension, while still supporting standard management interfaces and protocols(e.g. NetFlow, sFlow, IPFIX, RSPAN, CLI, LACP, 802.1ag). In addition, it isdesigned to support distribution across multiple physical servers similar toVMware's vNetwork distributed vswitch or Cisco's Nexus 1000V.
- VerneMQ - VerneMQ is a high-performance, distributedMQTT message broker. It scales horizontally and vertically on commodityhardware to support a high number of concurrent publishers and consumers whilemaintaining low latency and fault tolerance. VerneMQ is the reliable messagehub for your IoT platform or smart products.
- aBeacon Data - A large-scalerepository composed of BLE sensing, location trace, and manual report data,including 31,131 couriers at 2,466 merchant locations in one month.
- AI Benchmark - The performance and comparison ofall chipsets from Qualcomm, HiSilicon, Samsung, MediaTek and Unisoc that areproviding hardware acceleration for AI inference.
- CloudSuite - CloudSuite is a benchmark suite forcloud services. The third release consists of eight applications that havebeen selected based on their popularity in today’s datacenters. The benchmarksare based on real-world software stacks and represent real-world setups.
- Edge AIBench - EdgeAIBench is a benchmark suite for end-to-end edge computing including fourtypical application scenarios: ICU Patient Monitor, Surveillance Camera, SmartHome, and Autonomous Vehicle, which consider the complexity of all edgecomputing AI scenarios. In addition, Edge AIBench provides an end-to-endapplication benchmarking framework, including train, validate and inferencestages.
- EUA Datasets - This repositorymaintains a set of EUA datasets which we collected from real-world datasources. The datasets are publicly released to facilitate research in EdgeComputing. All the data is in Australia region which contains edge serverlocations and user location.
- Rocketfuel - :Traceroutes were sourced from 800 vantage points hosted by nearly 300traceroute web servers. They started by mapping 10 ISPs, in Europe, Australiaand the United States. In that process, they constructed a database of over 50thousand IP addresses representing 45 thousand routers in 537 POPs connectedby 80 thousand links.
- netem - netem providesNetwork Emulation functionality for testing protocols by emulating theproperties of wide area networks. The current version emulates variable delay,loss, duplication and re-ordering.
- Sigar - The Sigar APIprovides a portable interface for gathering system information such as:
- Eman's Edge Computing System For AI Applications - :Highly accurate AI applications, particularly for Computer Vision requiresextensive memory and computational power. Eman's Edge Computing Systemorchestrates resource limited devices that requires using resource heavy AIalgorithms so that a system can still achieve ideal system performances. Thisis done through Edge's monitoring modules that determines how to adjust thesystem so the system can operate efficiently based on its environmentsconditions.
- Apache TVM - Apache TVM is an open source machinelearning compiler framework for CPUs, GPUs, and machine learning accelerators.It aims to enable machine learning engineers to optimize and run computationsefficiently on any hardware backend.
- Core ML - Core ML optimizeson-device performance by leveraging the CPU, GPU, and Neural Engine whileminimizing its memory footprint and power consumption. Running a modelstrictly on the user’s device removes any need for a network connection, whichhelps keep the user’s data private and your app responsive.
- DeepStack - DeepStack is an Open-Source AI APIengine that serves pre-built models and custom models on multiple edge deviceslocally or on your private cloud.
- GNN-RL pipleline - GNN-RL pipleline is atoolkit to help users extract topology information through reinforcementlearning task (e.g., topology-aware neural network compression and jobscheduling). GNN-RL contains two major part -- graph neural network (GNN) andreinforcement learning (RL). It requires your research object is a graph orbeen modeled as a graph, the reinforcement learning take the graph asenvironment states and produces actions by using a GNN-based policy network.
-
ML Kit
- ML Kit 19 is a mobile SDKframework introduced by Google. It uses Google's cloud vision APIs, mobilevision APIs, and TensorFlow Lite to perform tasks like text recognition, imagelabeling, and smart reply. Curukogluall et al. tested ML Kit APIs for imagerecognition, bar-code scanning, and text recognition on an Android device andreported that these APIs recognize different types of test objects such as teacup, water glass, remote controller, and computer mouse successfully. src
from: 10.1145/3469029 - NVIDIA TensorRT - NVIDIA® TensorRT™ isan SDK for high-performance deep learning inference. It includes a deeplearning inference optimizer and runtime that delivers low latency and highthroughput for deep learning inference applications.
- PyTorch Mobile - The PyTorch Mobile runtimebeta release allows you to seamlessly go from training a model to deployingit, while staying entirely within the PyTorch ecosystem. It provides anend-to-end workflow that simplifies the research to production environment formobile devices. In addition, it paves the way for privacy-preserving featuresvia federated learning techniques.
- Komondor - Komondor is a wireless networksimulator that includes novel mechanisms for next-generation WLANs, such asdynamic channel bonding or enhanced spatial reuse. One of the main purposes ofKomondor is to simulate the behavior of IEEE 802.11ax-2019 networks, anamendment designed to boost spectral efficiency in dense deployments.
- POX - POX is a networking software platformwritten in Python. POX started life as an OpenFlow controller, but can nowalso function as an OpenFlow switch, and can be useful for writing networkingsoftware in general. An useful link of how to simulate a SDN can check thelink: http://www.brianlinkletter.com/using-the-pox-sdn-controller/
-
RICE
- A unified approach to remotefunction invocation in ICN(Naming Data Network
Platform) that exploits theattractive ICN properties of name-based routing, receiver-driven low andcongestion control, low balance, and object-oriented security while presentinga natural programming model to the application developer. - Wonder Shaper - Wonder Shaper is ascript that allows the user to limit the bandwidth of one or more networkadapters. It does so by using iproute's tc command, but greatly simplifies itsoperation.
- DeFog - DeFog, a first Fog benchmarkingsuite to: (i) alleviate the burden of Fog benchmarking by using a standardmethodology, and (ii) facilitate the understanding of the target platform bycollecting a catalogue of relevant metrics for a set of benchmarks.
- MLPerf Inference Benchmark Suite - :MLPerf Inference is a benchmark suite for measuring how fast systems can runmodels in a variety of deployment scenarios.
- ASSOLO - ASSOLO is a new active probingtool for estimating available bandwidth based on the concept of "self-inducedcongestion". ASSOLO features a new probing traffic profile called REACH(Reflected ExponentiAl Chirp), which tests a wide range of rates being moreaccurate in the center of the probing interval. Moreover, the tool runs insidea real-time operating system and uses some de-noising techniques to improvethe measurement process.
-
nmon
- This systems administrator,tuner, benchmark tool gives you a huge amount of important performanceinformation in one go. Many helpful tools for data analysis can be also foundon section Data
Analysis. - Adlik - Adlik [ædlik] is an end-to-endoptimizing framework for deep learning models. The goal of Adlik is toaccelerate deep learning inference process both on cloud and embeddedenvironment.
- AI Model Efficiency Toolkit (AIMET) - :AIMET isa library that provides advanced model quantization and compression techniquesfor trained neural network models. It provides features that have been provento improve run-time performance of deep learning neural network models withlower compute and memory requirements and minimal impact to task accuracy.
- AutoDiCE - A tool featuring automatedsplitting of a CNN model into a set of submodels and automated code generationfor distributed and collaborative execution of these submodels on multiple,possibly heterogeneous, edge devices. AutoDiCE is the first fully automatedtool for distributed CNN inference over multiple resource-constrained devicesat the Edge.
- BerryNet - This project turns edge devicessuch as Raspberry Pi into an intelligent gateway with deep learning running onit. No internet connection is required, everything is done locally on the edgedevice itself. Further, multiple edge devices can create a distributed AIoTnetwork.
- Bi-Real Net - Compared with thestandard 1-bit CNNs, Bi-Real net utilizes a simple short-cut to significantlyenhance the representational capability. Further, an advanced trainingalgorithm is specifically designed for training 1-bit CNNs (including Bi-Realnet), including a tighter approximation of the derivative of the sign functionwith respect the activation, the magnitude-aware gradient with respect to theweight, as well as a novel initialization.
- BMXNet - Apache MXNet is a deep learningframework designed for both efficiency and flexibility. It allows you to mixsymbolic and imperative programming to maximize efficiency and productivity.At its core, MXNet contains a dynamic dependency scheduler that automaticallyparallelizes both symbolic and imperative operations on the fly. A graphoptimization layer on top of that makes symbolic execution fast and memoryefficient. MXNet is portable and lightweight, scalable to many GPUs andmachines.
- BranchyNet - Fast inference via earlyexiting from deep neural networks. The architecture allows prediction resultsfor a large portion of test samples to exit the network early via thesebranches when samples can already be inferred with high confidence.
- Communication-Aware DNN Pruning (CaP) - Anovel distributed inference framework for distributing DNN computations acrossa physical network. Departing from conventional pruning methods, CaP takes thephysical network topology into consideration and produces DNNs that arecommunication-aware, designed for both accurate and fast execution over such adistributed deployment.
- Compute Library - The ComputeLibrary is a collection of low-level machine learning functions optimized forArm® Cortex®-A and Arm® Mali™ GPUs architectures.
- Condensa - CONDENSA allows developers todesign strategies for compressing DL models, resulting significant reductionin both the memory footprint and execution time. It uses a Bayesianoptimization-based method to find compression hyperparameters automatically byexploiting an L-C optimizer to recover the accuracy loss during compression.They reported a 2.22 times runtime improvement over an uncompressed VGG-19 onthe CIFAR-10 dataset with up to 65 times memory reduction. src from: 10.1145/3469029
- daBNN - daBNN is an open-source fastinference framework developed by Zhang et al., which can implement BinaryNeural Networks on ARM devices [161]. An upgraded bit-packing scheme andbinary direct convolution have been used in this framework to shrink the costof convolution and speed up inference. This framework is written in C++ andARM assembly and has Java support for the Android package. This fast frameworkcan be 6 times faster than BMXNet onBi-Real Net. src from: 10.1145/3469029
- DDNN - distributed deep neural networks(DDNNs) over distributed computing hierarchies, consisting of the cloud, theedge (fog) and end devices. While being able to accommodate inference of adeep neural network (DNN) in the cloud, a DDNN also allows fast and localizedinference using shallow portions of the neural network at the edge and enddevices. Due to its distributed nature, DDNNs enhance data privacy and systemfault tolerance for DNN applications. When supported by a scalable distributedcomputing hierarchy, a DDNN can scale up in neural network size and scale outin geographical span.
- DeepIoT - DeepIoT is a framework thatshrinks a neural network into smaller dense matrices but keeps the performanceof the algorithm almost the same. This framework finds the minimumnumber of filters and dimensions required by each layer and reduces theredundancy of that layer. src from: 10.1145/3469029
- DeepThings - A framework for locallydistributed and adaptive CNN inference in resource-constrained IoT edgeclusters. DeepThings mainly consists of: 1). A Fused Tile Partitioning (FTP)method for dividing convolutional layers into independently distributabletasks. FTP fuses layers and partitions them vertically in a grid fashion,which largely reduces communication and task migration overhead. 2). Adistributed work stealing runtime system for IoT clusters to adaptivelydistribute FTP partitions in dynamic application scenarios.
- Distiller - Distiller is anopen-source Python package for neural network compression research. Itprovides a PyTorch environment for prototyping and analyzing compressionalgorithms, such as sparsity-inducing methods and low-precision arithmetic.
- FATE - :FATE (Federated AI TechnologyEnabler) is an open-source project initiated by Webank's AI Department toprovide a secure computing framework to support the federated AI ecosystem. Itimplements secure computation protocols based on homomorphic encryption andmulti-party computation (MPC). It supports federated learning architecturesand secure computation of various machine learning algorithms, includinglogistic regression, tree-based algorithms, deep learning and transferlearning.
- FedProx - FedProx aims to solve two keychallenges that differentiate it from traditional distributed optimization:(1) significant variability in terms of the systems characteristics on eachdevice in the network (systems heterogeneity), and (2) non-identicallydistributed data across the network (statistical heterogeneity).
- Ianvs - Ianvs is a distributed synergy AIbenchmarking project incubated in KubeEdge SIG AI. Ianvs aims to test theperformance of distributed synergy AI solutions following recognizedstandards, in order to facilitate more efficient and effective development.More detailedly, Ianvs prepares not only test cases with datasets andcorresponding algorithms, but also benchmarking tools including simulation andhyper-parameter searching. Ianvs also reveals best practices for developersand end users with presentation tools including leaderboards and test reports.
- KitNET - KitNET is an online,unsupervised, and efficient anomaly detector. A Kitsune, in Japanese folklore,is a mythical fox-like creature that has a number of tails, can mimicdifferent forms, and whose strength increases with experience. Similarly,Kit-NET has an ensemble of small neural networks (autoencoders), which aretrained to mimic (reconstruct) network traffic patterns, and whose performanceincrementally improves overtime.
- MACE - MACE (Mobile AI Compute Engine) is adeep learning inference framework optimized for mobile heterogeneous computingplatforms. MACE provides tools and documents to help users to deploy deeplearning models to mobile phones, tablets, personal computers and IoT devices.
- MegEngine - MegEngine is a fast,scalable and easy-to-use deep learning framework, with auto-differentiation.
- MLC LLM - In recent years, there has beenremarkable progress in generative artificial intelligence (AI) and largelanguage models (LLMs), which are becoming increasingly prevalent. Thanks toopen-source initiatives, it is now possible to develop personal AI assistantsusing open-sourced models. However, LLMs tend to be resource-intensive andcomputationally demanding. To create a scalable service, developers may needto rely on powerful clusters and expensive hardware to run model inference.Additionally, deploying LLMs presents several challenges, such as theirever-evolving model innovation, memory constraints, and the need for potentialoptimization techniques.
- MNN - MNN is a highly efficient andlightweight deep learning framework. It supports inference and training ofdeep learning models, and has industry leading performance for inference andtraining on-device. At present, MNN has been integrated in more than 20 appsof Alibaba Inc, such as Taobao, Tmall, Youku, Dingtalk, Xianyu and etc.,covering more than 70 usage scenarios such as live broadcast, short videocapture, search recommendation, product searching by image, interactivemarketing, equity distribution, security risk control. In addition, MNN isalso used on embedded devices, such as IoT.
- Model Compression Toolkit (MCT) - :Model Compression Toolkit (MCT) is an open source project for neural networkmodel optimization under efficient, constrained hardware. This projectprovides researchers, developers, and engineers advanced quantization andcompression tools for deploying state-of-the-art neural networks.
- MQBench - MQBench is an open-source modelquantization toolkit based on PyTorch fx. The envision of MQBench is toprovide: 1) SOTA Algorithms. With MQBench, the hardware vendors andresearchers can benefit from the latest research progress in academia. 2)Powerful Toolkits. With the toolkit, quantization node can be inserted to theoriginal PyTorch module automatically with respect to the specific hardware.After training, the quantized model can be smoothly converted to the formatthat can inference on the real device.
- ncnn - ncnn is a high-performance neuralnetwork inference computing framework optimized for mobile platforms. ncnn isdeeply considerate about deployment and uses on mobile phones from thebeginning of design. ncnn does not have third party dependencies. it iscross-platform, and runs faster than all known open source frameworks onmobile phone cpu. Developers can easily deploy deep learning algorithm modelsto the mobile platform by using efficient ncnn implementation, createintelligent APPs, and bring the artificial intelligence to your fingertips.ncnn is currently being used in many Tencent applications, such as QQ, Qzone,WeChat, Pitu and so on.
- Neurosurgeon - :A lightweight schedulerto automatically partition DNN computation between mobile devices anddatacenters at the granularity of neural network layers.
- nn-Meter - :nn-Meter is a novel andefficient system to accurately predict the inference latency of DNN models ondiverse edge devices. The key idea is dividing a whole model inference intokernels, i.e., the execution units of fused operators on a device, and conductkernel-level prediction.
- nndeploy - :nndeploy is an end-to-end deployment framework for models. Centered aroundmulti-end inference and model deployment based on directed acyclic graph, itis committed to providing users with a cross-platform, simple, easy-to-use,and high-performance model deployment experience.
- ns3-ai - :This module does notprovide any AI algorithms or rely on any frameworks but instead is providinga Python module that enables AI interconnect, so the AI framework needs to beseparately installed. You only need to clone or download this work, thenimport the Python modules, you could use this work to exchange databetween ns-3 and your AI algorithms.
- ns3-gym - OpenAI Gym is a toolkit forreinforcement learning (RL) widely used in research. The network simulatorns–3 is the de-facto standard for academic and industry studies in the areasof networking protocols and communication technologies. ns3-gym is a frameworkthat integrates both OpenAI Gym and ns-3 in order to encourage usage of RL innetworking research.
- Once for All - Train One Networkand Specialize it for Efficient Deployment.
- ONNX Runtime - ONNX Runtime is across-platform inference and training machine-learning accelerator. It canenable faster customer experiences and lower costs, supporting models fromdeep learning frameworks such as PyTorch and TensorFlow/Keras as well asclassical machine learning libraries such as scikit-learn, LightGBM, XGBoost,etc. ONNX Runtime is compatible with different hardware, drivers, andoperating systems, and provides optimal performance by leveraging hardwareaccelerators where applicable alongside graph optimizations and transforms.
- Paddle-Lite - Paddle Lite is anupdated version of Paddle-Mobile, an open-open source deep learning frameworkdesigned to make it easy to perform inference on mobile, embeded, and IoTdevices. It is compatible with PaddlePaddle and pre-trained models from othersources.
-
Pocket
- A new approach for serving MLapplications in settings like the edge, based on a shared ML runtime backendas a service and lightweight ML application pocket containers. Key torealizing Pocket is use of lightweight IPC, support for cross-clientisolation, and a novel resource amplification method which inlines resourcereallocation with IPC. The latter ensures just-in-time assignment of thelimited edge resources where they're most needed, thereby reducing contentioneffects and boosting overall performance and efficiency. Pocket: ML Serving
from the Edge - SparseML - SparseML is anopen-source model optimization toolkit that enables you to createinference-optimized sparse models using pruning, quantization, anddistillation algorithms. Models optimized with SparseML can then be exportedto the ONNX and deployed with DeepSparse for GPU-class performance on CPUhardware.
- SparseZoo - SparseZoo is aconstantly-growing repository of sparsified (pruned and pruned-quantized)models with matching sparsification recipes for neural networks. It simplifiesand accelerates your time-to-value in building performant deep learning modelswith a collection of inference-optimized models and recipes to prototype from.
- Tengine - Tengine is developed by OPEN AILAB. This project meet the demand of fast and efficient deployment of deeplearning neural network models on embedded devices. In order to achievecross-platform deployment in many AIoT applications, this project is based onthe original Tengine project using C language for reconstruction, and deepframe tailoring for the characteristics of limited embedded device resources.Also, it adopts a completely separated front-end/back-end design, which makesit possible to be transplanted and deployed onto CPU, GPU, NPU and otherheterogeneous computing units rapidly, conveniently.
- TensorFlow Lite - TensorFlow Lite is anopen-source deep learning framework to run TensorFlow models on-device. If youare new to TensorFlow Lite, we recommend that you first explore thepre-trained models and run the example apps below on a real device to see whatTensorFlow Lite can do.
- TNN - TNN: A high-performance, lightweightneural network inference framework open sourced by Tencent Youtu Lab. It alsohas many outstanding advantages such as cross-platform, high performance,model compression, and code tailoring. The TNN framework further strengthensthe support and performance optimization of mobile devices on the basis of theoriginal Rapidnet and ncnn frameworks. At the same time, it refers to the highperformance and good scalability characteristics of the industry's mainstreamopen source frameworks, and expands the support for X86 and NV GPUs. On themobile phone, TNN has been used by many applications such as mobile QQ,weishi, and Pitu. As a basic acceleration framework for Tencent Cloud AI, TNNhas provided acceleration support for the implementation of many businesses.Everyone is welcome to participate in the collaborative construction topromote the further improvement of the TNN inference framework.
-
Torch-Pruning (TP)
- TP enablesstructural pruning for a wide range of deep neural networks, including Large
Language Models
(LLMs),Segment Anything Model (SAM), Diffusion
Models, Vision
Transformers,ConvNext,Yolov7, yolov8, Swin
Transformers,BERT, FasterRCNN, SSD,ResNe(X)t, DenseNet, RegNet, DeepLab, etc. Different fromtorch.nn.utils.prunethat zeroizes parameters through masking, Torch-Pruning deploys an algorithmcalled**DepGraph**to remove parameters physically. - uTensor - uTensor is an extremelylight-weight machine learning inference framework built on Tensorflow andoptimized for Arm targets. It consists of a runtime library and an offlinetool that handles most of the model translation work. This repo holds the coreruntime and some example implementations of operators, memorymanagers/schedulers, and more, and the size of the core runtime is only ~2KB!
- AI at the edge - A curated list ofresources for embedded AI.
- Explore Edge Computing - :Discover & find a curated list of popular & new libraries, top authors,trending project kits, discussions, tutorials & learning resources on kandi.
- MEC Ecosystem - :Provides information very much related to the work of the ETSI ISG MECDeployment and ECOsystem DEvelopment (DECODE) Working Group, whose aim is toaccelerate the development of the MEC ecosystem
- Shimmy - Self-hosted AI inferenceserver with OpenAI API compatibility, designed for edge deployment. Supportsmultiple model formats (GGUF, Core ML, etc.) with built-in streaming andmultimodal capabilities. Lightweight and fast for resource-constrained edgeenvironments.
- Caffe2 - :Caffe2 is alightweight, modular, and scalable deep learning framework. Building on theoriginal Caffe, Caffe2 is designed with expression, speed, and modularity inmind.
- MindSpore Lite - MindSpore Lite is anultra-fast, intelligent, and simplified AI engine that enables intelligentapplications in all scenarios, provides E2E solutions for users, and helpsusers enable AI capabilities.
- ClusterData - Alibaba ClusterData isa collection of production cluster traces for cloud scheduling, resourcemanagement, and data center systems research.
- Genesis 2 - CPU-only neural network for edge deployment. Cascade MoE with 18ms inference, 3.5GB model, no GPU required. Ideal for edge AI applications.
-
Mosquitto
- Eclipse Mosquitto is an open source(EPL/EDL licensed) message broker that implements the MQTT protocol versions5.0, 3.1.1 and 3.1. Mosquitto is lightweight and is suitable for use on alldevices from low power single board computers to full servers. The MQTTprotocol provides a lightweight method of carrying out messaging using apublish/subscribe model. This makes it suitable for Internet of Thingsmessaging such as with low power sensors or mobile devices such as phones,embedded computers or microcontrollers. The Mosquitto project also provides aC library for implementing MQTT clients, and the very popular
mosquitto_pubandmosquitto_subcommand line MQTT clients. - GT-ITM: Georgia Tech Internetwork TopologyModels - This is a collection ofroutines to generate and analyze graphs using a wide variety of models forinternetwork topology. The graphs are generated in Don Knuth's SGB format; aroutine is provided to convert to an alternative format that may be easier toparse.
- Huawei - Dataset for Network AI Scheduling TechnologyResearch (in Chinese) - :Modeling computational service scenarios for Network AI, providing users withstrategy algorithms for task scheduling research.
- rk-bench - A standardized benchmark leaderboard for public models on Rockchip boards (RK3588), measuring latency (p50/p95) with the NPU runtime's own hardware counters and reporting per-operator CPU fallback from PERF_DETAIL; includes FP/INT8 comparisons and core-mask studies.
- Network ModelingDatasets - This repositoryprovides datasets generated using a custom OMNeT++ simulator, covering diversenetwork topologies, routing configurations and traffic patterns. Each sampleincludes precise end-to-end performance metrics, such asper-source–destination packet-level delay, jitter and loss, with detailedinstructions available in each dataset’s root directory. These datasets areintended to facilitate the design, training and evaluation of ML-based networkmodels and to support fair benchmarking against prior state-of-the-artapproaches on a common dataset.
- Neural Network AcceleratorComparison - :A comparison of AI hardware accelerators among different platform.
- The CAIDA Anonymized Internet Traces 2015Dataset - CAIDA'spassive traces dataset contains traces collected from high-speed monitors on acommercial backbone link. The data collection started in April 2008 and endedin January 2019. These data are useful for research on the characteristics ofInternet traffic, including application breakdown, security events, geographicand topological distribution, flow volume and duration.
- tinyMLPerf Deep Learning Benchmarks for EmbeddedDevices - The goal of TinyMLPerf is toprovide a representative set of deep neural nets and benchmarking code tocompare performance between embedded devices. Embedded devices includemicrocontrollers, DSPs, and tiny NN accelerators. These devices typically runat between 10MHz and 250MHz, and can perform inference using less then 50mW ofpower.
- https://commonbox.vivo.xyz/s/rUeNxBRZfKL
- https://commonbox.vivo.xyz/s/dHMSmyouTeC
- FRACTAL - A decentralizedframework for federated learning, distributed edge computing, and training modelsdirectly on Android and edge devices.
- Cloud Computing and Distributed Systems (CLOUDS)Laboratory - The Cloud Computing andDistributed Systems (CLOUDS) Laboratory, formerly GRIDS Lab, is a softwareresearch and development group within the School of Computing and InformationSystems at the University of Melbourne, Australia. The CLOUDS Lab is activelyengaged in the design and development of next-generation computing systems andapplications that aggregate or lease services of distributed resourcesdepending on their availability, capability, performance, cost , and users'quality-of-service requirements. The lab is working towards realising thisvision through its two flagship projects: Gridbus and Cloudbus.
- Edge Computing & Internet OfThings - A qualitative compilationof production-ready edge computing projects with a focus on IoT based DataEngineering.
Uncategorized
Uncategorized
- Cooja - CoojaSimulator is a network simulator specifically designed for Wireless SensorNetworks.
- gem5 - The gem5 simulator is a modular platform forcomputer-system architecture research, encompassing system-level architectureas well as processor microarchitecture.
- lightMANO - A lightweightdistributed service orchestrator designed for resource constrainedmulti-access edge computing environments.
- Mini-NDN - :Mini-NDN is a lightweight networkingemulation tool that enables testing, experimentation, and research on the NDNplatform. It was initially based onMini-CCNx which was a fork of Mininet.Mini-NDN uses the NDN libraries, NFD, NLSR, and tools released by the NDNproject to emulate an NDN network on a single system.
- RECAP Simulator Framework - The RECAPSimulation Framework is an open source simulation framework designed tosupport experimentation of infrastructure with different description modelsfor workload, user distribution, network topology, and (physical and virtual)resource placement. It is able to simulate different use cases includingVirtual Network Functions (VNFs), Elasticsearch, smart cities, and virtualContent Delivery Networks (vCDN). The output of the RECAP Simulation Frameworkdepends on the use case, but in general, it is a set of metrics predefined bythe user such as bandwidth, resource consumption (CPU, memory, storage),network delay, energy consumption, active number of VMs, cache hits and cachemisses.
- SUMO - ITS - "Simulation ofUrban MObility", or "SUMO" for short, is an open source, microscopic,multi-modal traffic simulation. It allows to simulate how a given trafficdemand which consists of single vehicles moves through a given road network.The simulation allows to address a large set of traffic management topics. Itis purely microscopic: each vehicle is modelled explicitly, has an own route,and moves individually through the network. Simulations are deterministic bydefault but there are various options for introducing randomness.
- SimFlex - SimFlex is proceeding along twosynergistic fronts: (1) Flexus is a powerful and flexible simulator frameworkthat allows full-system simulation that relies heavily on well-definedcomponent interface models to facilitate both model integration andcompile-time simulator optimization. (2) SMARTS applies rigorous statisticalsampling theory to reduce simulation turnaround by several orders ofmagnitude, while achieving high accuracy and confidence in estimates.
- VENTOS - VENTOS is an integrated C++simulator for studying vehicular traffic flows, collaborative driving, andinteractions between vehicles and infrastructure through DSRC-enabled wirelesscommunication capability.
- VNS - VNS is a simulationframework that completely integrates the mobility and network components in atransparent and efficient way, reducing the overhead of communication andsynchronization between different simulators. VNS provides bi-directionallyinteraction between a microscopic mobility model and network simulators suchas OMNeT++.
- SimuLTE - SimuLTE is an innovative simulation toolenabling complex system level performance-evaluation of LTE and LTE Advancednetworks (3GPP Release 8 and beyond) for the OMNeT++ framework.
- CausalSim - CausalSimis a causal framework for unbiased trace-driven simulation. Currenttrace-driven simulators assume that the interventions being simulated (e.g., anew algorithm) would not affect the validity of the traces. However,real-world traces are often biased by the choices algorithms make during tracecollection, and hence replaying traces under an intervention may lead toincorrect results. CausalSim addresses this challenge by learning a causalmodel of the system dynamics and latent factors capturing the underlyingsystem conditions during trace collection. It learns these models using aninitial randomized control trial (RCT) under a fixed set of algorithms, andthen applies them to remove biases from trace data when simulating newalgorithms.
- CloudSim - CloudSim goal is to providea generalized and extensible simulation framework that enables modeling,simulation, and experimentation of emerging Cloud computing infrastructuresand application services, allowing its users to focus on specific systemdesign issues that they want to investigate, without getting concerned aboutthe low level details related to Cloud-based infrastructures and services.
- CloudSim Express - :CloudSim Express introduces a human-readable scripting approach to definesimulation scenarios, a standard way of writing extensions, and a simplifieddeployment approach. The script-based approach significantly reducescode-based implementations. In the cases of simple examples, it completelyeliminates the need to implement Java code.
- CloudSim Plus - A modern,up-to-date, full-featured and fully documented Java 17 simulation framework.It's easy to use and extend, enabling modeling, simulation, andexperimentation of Cloud computing infrastructures and application services.It allows developers to focus on specific system design issues to beinvestigated, without concerning the low-level details related to Cloud-basedinfrastructures and services.
- CloudSim+ - Py4j gateway - :py4j gateway for Cloud Sim Plus framework.
- CloudSimSDN - An SDN extension ofCloudSim project to simulate SDN features in the context of a cloud datacenter. CloudSimSDN supports calculating power consumption by both hosts andswitches. For instance, network-aware VM placement policies can be evaluatedusing CloudSimSDN.
- CORE - The Common Open Research Emulator(CORE) is a tool for emulating networks on one or more machines. You canconnect these emulated networks to live networks. CORE consists of a GUI fordrawing topologies of lightweight virtual machines, and Python modules forscripting network emulation.
- EasiEI - EasiEI simulates variouscomputing, communication, and storage resources in edge computing, andprovides basic task scheduling algorithms to support the simulation of edgecomputing scenarios.
- ECSNeT++ - ECSNeT++ is a simulationtoolkit for simulating the execution of Distributed Stream Processingapplications on Edge anc Cloud Computing environments. ECSNeT++ is implementedusing the OMNeT++ and the INET framework.
- EdgeCloudSim - EdgeCloudSim:An Environment for Performance Evaluation of Edge Computing Systems.
- EdgeSim - :An open-source simulatorof edge computing and caching. This simulator system contains three models:Content, Base Station and User.
- EmuFog - EmuFog helps to test fog computingapplications more efficiently. Instead of actual deploying large networktopologies with your application to test, EmuFog helps to generate networksthat can be emulated easily with MaxiNet, adistributed version of the popular Mininet. Thisprovides more realistic results than simulations and is cheaper and fasterthan real deployments. As an input EmuFog supports generated topologies fromBRITE or measured real world topologies fromCaida. In those networks EmuFog places fog nodesefficiently based on user defined constrains such as network latencythresholds or resource constraints. Applications for clients and fog nodes canbe anything shipped in a Docker container.
- EPCSAC - The EPCSAC (Extensible Platformfor Cloud Scheduling Algorithm Comparison) is an online open-source platformdeveloped to help researchers compare scheduling algorithms to allocate tasksinto virtual machines inside cloud infrastructures.
- FAAP-Simulator - Applicationallocation in the area of fog computing and targets fog networks in industrialenvironments. With the simulator, it is possible to generate, run and evaluateinstances of the Fog Application Allocation Problem denoted as FAAP. Thesimulator is completely implemented in the scala programming language.
- faas-sim - Faas-sim is a powerfultrace-driven simulation framework to simulate container-basedfunction-as-a-service platforms. It can be used to develop, and evaluate theperformance of operational strategies for such systems, like scheduling,autoscaling, load balancing, and others. faas-sim was developed at theDistributed Systems Group at TU Wien as part of a larger research effortsurrounding serverless edge computing systems.
- Fogbed - Fogbed is a framework that extendsthe Mininet emulator to create fog testbeds in virtualized environments. Usinga desktop approach, Fogbed enables the deployment of virtual fog nodes asDocker containers under different network configurations. The Fogbed APIprovides functionality to add, connect and remove containers dynamically fromthe network topology. These features allow for the emulation of real worldcloud and fog infrastructures in which it is possible to start and stopcompute instances at any point in time. Also, it is possible to change atruntime resource limitations for a container, such as CPU time and memoryavailable.
- Fogify - Fogify is an emulationFramework easing the modeling, deployment and experimentation of fog testbeds.Fogify provides a toolset to: model complex fog topologies comprised ofheterogeneous resources, network capabilities and QoS criteria; deploy themodelled configuration and services using popular containerizedinfrastructure-as-code descriptions to a cloud or local environment;experiment, measure and evaluate the deployment by injecting faults andadapting the configuration at runtime to test different "what-if" scenariosthat reveal the limitations of a service before introduced to the public.
- FogNetSim++ - FogNetSim++extends OMNeT++, which is a well known framework for building networksimulators, to model all these aspects. Moreover, it includes popularcommunication protocols for simulation, such as TCP, UDP, MQTT, and CoAP.Furthermore, FogNetSim++ models several other aspects, such as energyconsumption, pricing, mobility, and handoff mechanisms.
- FogTorchPI - FogTorchΠ is anopen source prototype, developed in Java, based on a model for Fog computinginfrastructures and applications. 1), It takes into account non-functionalparameters within the model (i.e., hardware, software, latency and bandwidth)to determine, compare and contrast different eligible deployments of a givenapplication over a Fog infrastructure. 2), In the case of hardwarecapabilities, it considers CPU cores, RAM and storage available at a givennode or required by a given software component. 3), Software capabilities arerepresented by a list of software names (operating system, programminglanguages, frameworks etc). 4), It considers latency, and both download andupload bandwidths as QoS attributes. Latency is measured in milliseconds (ms),while bandwidth is given in Megabits per second (Mbps). Note: Outputs ofFogTorchPI can be the input of iFogSim.
- iFogSim - The iFogSim Toolkit forModeling and Simulation of Resource Management Techniques in Internet ofThings, Edge and Fog Computing Environments.
- IoTSim-Edge - IoTSim captures thebehavior of heterogeneous IoT and edge computing infrastructure and allowsusers to test their infrastructure and framework in an easy and configurablemanner. IoTSim-Edge extends the capability of CloudSim to incorporate thedifferent features of edge and IoT devices.
- IoTSim-Osmosis - IoTSim-Osmosisis a simulation framework that supports the testing and validation of osmoticcomputing applications. In particular, it enables a unified modelling andsimulation of complex IoT applications over heterogeneous edge-cloud SDN-awareenvironments.DOI:10.1016/j.sysarc.2020.101956
- Kathará - Kathará (from theGreek Καθαρά, purely) is a lightweight network emulation system based onDocker containers. It can be really helpful in showing interactivedemos/lessons, testing production networks in a sandbox environment, ordeveloping new network protocols.
- LEAF - A simulator for LargeEnergy-Aware Fog computing environments, based on CloudSim Plus. LEAF enablesenergy consumption modeling of distributed, heterogeneous, andresource-constrained infrastructure that executes complex application graphs.Its Python version.
- MaxiNet - MaxiNet extends the famousMininet emulation environment to span the emulation across several physicalmachines. This allows to emulate very large SDN networks.
-
MobFogSim
- An extension which cansupport for Mobility of iFogSim. MobFogSim can model more generalisedaspects related to device mobility and VM/container migration in the fog,e.g., user position and speed, connection handoff, migration policies andstrategies, to name a few. More details can be found in MobFogSim: Simulation
of Mobility and Migration for Fog
Computing. - Mininet-WiFi - Emulator forSoftware-Defined Wireless Networks.
- MobEmu - An opportunistic networkemulator that can run a user-created routing or dissemination algorithm on adesired mobility trace or synthetic model.
- NDN4IVC - An open-sourcesimulator/framework that facilitates testing of more realistic VANETapplications via the NDN stack. This project utilizes two popular simulatorsfor VANET simulation: the NS3 network simulator with the ndnSIM module andSUMO, a simulator of urban mobility. NDN4IVC allows real-time bidirectionalcommunication between SUMO and NS3 to support more data about road traffic andvehicular mobility. The framework can model the impact of vehicular networkson road traffic and investigate complex interactions between the two domains.
- NFaaS - Enabling to migrate functionscloser to the user in ICN environment without a global view of the network.
- NoSSim - Network-of-Systems Simulator(NoSSim) is a network/system co-simulation framework that combines a fast andaccurate host-compiled full system simulator with a standard, reconfigurableOMNeT++ network simulation backplane. Detailed models of network interfacesand protocol stacks (lwIP) are integrated into host-compiled system and OSmodels to allow accurately capturing network and system interactions.
- OMNeT++ - OMNeT++ is a public-source,component-based, modular and open-architecture simulation environment withstrong GUI support and an embeddable simulation kernel. Its primaryapplication area is the simulation of communication networks, but it has beensuccessfully used in other areas like the simulation of IT systems, queueingnetworks, hardware architectures and business processes as well.
- Open-Simulator - Open-simulator isa cluster simulator for Kubernetes. With the simulation capability ofOpen-Simulator, users can create a fake Kubernetes cluster and deployworkloads on it. Open-Simulator will simulate the kube-controller-manager tocreate pods for the workloads, and simulate the kube-scheduler to assign podsto the appropriate nodes.
- ECHOES - ECHOES stands for "Edge andCloud Hybrid Optimization Environment Simulator", which is a free andopen-source tool to simulate an Edge/Cloud-Hybrid network topology,in order to optimise and test methods for offloading Tasks from user's deviceto edge-server or cloud-server.
- EdgeSimPy - EdgeSimPy models thelifecycle of containerized applications through several functionalabstractions that replicate the behavior of container runtimes like Docker. Inaddition, EdgeSimPy features a flexible input format that allows users todefine custom parameters for simulated entities out-of-the-box, extending thesimulator’s built-in capabilities without modifying its core features.
- Emu5GNet - This emulationenvironment has been created to allow researchers/industrials to develop anddeploy applications in complex 5G network architecture (SA and NSA) forenabling both emulations of the network and data processing in realistic E2Escenarios. Emu5GNet allows the execution of 5G RAN and Core using Mininet-Wifienvironment and containernet. The 5G RANand Core can be packed inside docker containers. Emu5GNet also allows theintegration of full ETSI NFV compliant infrastructure based on [vim-emu](https://github.com/containernet/vim-emu) platform. This last integration ofan enhancement of vim-emu enables the interconnection of Mininet-WifiAccess points and vim-emu datacenters and, allows the network functionsmigration across edge datacenters.
- EmuEdge - EmuEdge is a scalable, Hi-Fi,highly-automated real-world network emulator based on Xen/Linux Netns/OvS,supporting edge computing prototyping and general network emulation with bothcontainer/vm and can be hybridly combined with existing physicalinfrastructures.
- EdgeAISim - A Python-basedtoolkit for simulating and modelling AI models in edge computing environments.EdgeAISim extends EdgeSimPy andincorporates AI models like Multi-Armed Bandit and Deep Q-Networks to optimisepower usage and task migration. The toolkit outperforms baseline worst-fitalgorithms, showcasing its potential in sustainable edge computing bysignificantly reducing power consumption and enhancing task management forvarious large-scale scenarios.
- MEC-simulator - An extensionof the network simulator ns-3, implementing entities (e.g., base stations, MECHosts, etc.), interactions, and functionalities (e.g., orchestration of MECresources, distribution of tasks, etc.) standardized in the context ofETSI-MEC, while supporting multi-cell scenarios with mobile users anddifferent user load generator models.
- MintEDGE - MintEDGE is a flexible edgecomputing simulation framework that allows the configuration of variousaspects of the infrastructure and enables researchers to test novel energyoptimization strategies.
- NextGSim - NextGSimprovides a simulation environment focusing on resource management aspects inmobile edge computing scenarios. It consists of mainly two parts: RadioNetwork simulation and Edge Computing simulation. These two simulations areloosely connected to enable both independent and joint use.
- Artery - Artery enables V2X simulations basedon ETSI ITS-G5 protocols like GeoNetworking and BTP. Single vehicles can beequipped with multiple ITS-G5 services through Artery's middleware, which alsoprovides common Facilities for these services. Some basic services, such asCooperative Awareness (CAMs) and Decentralized Notification (DENMs), arealready included.
- iTETRIS - iTETRIS isaligned with the communication architecture defined by ETSI for IntelligentTransport Systems (ITS). The standard specifications concern a communicationsystem designed for various types of traffic applications which can useseveral coexistent communication technologies. The architecture assumes threedifferent actors communicating in an ITS scenario, each representing a givensubsystem: vehicle, roadside and central subsystems.
- MARSSx86 - MARSSx86 (MARSSfor short) is a tool for cycle accurate full system simulation of the x86-64architecture, specifically multicore implementations.
-
MinNet
- Mininet creates a realistic virtual network,running real kernel, switch and application code, on a single machine (VM,cloud or native), in seconds, with a single command:
sudo mn. Mininet is alsoa great way to develop, share, and experiment with OpenFlow andSoftware-Defined Networking systems. - pFogSim - pFogSim (/p/-fôg-/sɪm/) is aplay off iFogSim; A simulator made to handle large-scale FOG networks with theHAFA Puddle Strategy to help evaluate the potential advantages/disadvantageswithin user-customizable scenarios.
- PureEdgeSim - a newsimulator based on CloudSim Plus that is designedto simulate Cloud, Edge, and Mist computing environments. It allows toevaluate the performance of resources management strategies in terms ofnetwork usage, latency, resources utilization, energy consumption, etc. andenables the simulation of several scenarios such as the Internet of Things(IoT), connected vehicles/VANETs/MANET, Mist computing environments (peer-topeer networks such as mobile devices Cloud), and mobile Edge computing.
- RaSim - RaSim (Raspberry-PiSimulator) is a software that simulates the motion and interaction of the edgethings in the IoT environment. The purpose of RaSim is to test through virtualthings before physically constructing things at the edge. Through simulations,it identifies information on structural problems, defects, pros/cons ofstrategies, resources usage to things of clusters built in edge computing.
- SatEdgeSim - A Toolkit forModeling and Simulation of Performance Evaluation in Satellite Edge ComputingEnvironments, which uses CloudSim Plus and PureEdgeSim as the underlyingsimulation framework.
- Shadow - Shadow is a discrete-event networksimulator that directly executes real application code, enabling you tosimulate distributed systems with thousands of network-connected processes inrealistic and scalable private network experiments using your laptop, desktop,or server running Linux.
- SimFaaS - This is a project done in PACSLab aiming to develop a performance simulator for serverless computingplatforms. Using this simulator, we can calculate Quality of Service (QoS)metrics like average response time, the average probability of cold start,average running servers (directly reflecting average cost), a histogram ofdifferent events, distribution of the number of servers throughout time, andmany other characteristics.
- SimGrid - SimGrid is a scientificinstrument to study the behavior of large-scale distributed systems such asGrids, Clouds, HPC or P2P systems. It can be used to evaluate heuristics,prototype applications or even assess legacy MPI applications.
- SimMobility - SimMobility is anintegrated mobility simulation platform that comprehensively simulates FutureMobility scenarios by integrating long, medium, and short-term travelbehavior. Various mobility-sensitive behavioral models are integrated withinthe state-of-the-art scalable simulators to predict the impact of mobilitydemands on transportation networks, intelligent transportation services andvehicular emissions. The platform simulates the effects of a portfolio oftechnology, policy and investment options under alternative future scenarios.SimMobility encompasses the modeling of millions of agents, includingpedestrians, drivers, phones, traffic lights, GPS, cars, buses, and trains,from second-by-second to year-by-year simulations and across countries.
- Simu5G - Simu5G is the evolution of thepopular SimuLTE 4G network simulator that incorporates5G New Radio access.
- SLEIPNIR - SLEIPNIR is a DAG schedulingsimulator focused on mobile cloud/edge/iot infrastructures.
- StarryNet - A novel experimentationframework that enables researchers to conveniently build credible and flexibleexperimental network environments (ENE) mimicking satellite dynamics andnetwork behaviors of large-scale integrated space and terrestrial networks(ISTNs) paper.
- Step-ONE - Simulated Testbed for Edge-FogProcesses based on the Opportunistic Network Environment simulator.
- SVL Simulator - An end-to-end autonomousvehicle simulation platform.
- The ONE - The ONE is a simulationenvironment that is capable of: 1) generating node movement using differentmovement models. 2) routing messages between nodes with various DTN routingalgorithms and sender and receiver types. 3) visualizing both mobility andmessage passing in real time in its graphical user interface. ONE can importmobility data from real-world traces or other mobility generators. It canalso produce a variety of reports from node movement to message passing andgeneral statistics.
- Veins - ITS - Veins is an open sourceframework for running vehicular network simulations. It is based on twowell-established simulators: OMNeT++, an event-based network simulator, andSUMO, a road traffic simulator. It extends these to offer a comprehensivesuite of models for IVC simulation.
- Veins LTE - ITS - Veins LTE is a simulator forheterogeneous vehicular networks. It provides fine-grained simulation ofvehicular networks based on IEEE 802.11p and TE.
- Vessim - A co-simulation testbed forcarbon-aware applications, applicable in edge computing environments. Vessimbuilds on the Mosaik framework, offering a high-level Python API to integraterenewable energy system simulators with real hardware andsoftware-in-the-loop capabilities.