sinofeng Goto Github PK

edge_cloud_computing

edge_computing

Peer-to-peer network of nodes receiving data from sensors spread over simulated city and sending elaborated data to the main server.

edgemw

Edge computing middleware distribution (Lightweight) - Highly in experimental stage

edgeworkloadstraces

edgy-controller

A monitoring-controller tool for dockerized services, inspired from control theory, used at the network edge. Code for paper "Where there is fire there is smoke: a scalable edge computing framework for early fire detection", https://www.mdpi.com/1424-8220/19/3/639/pdf

elephantflowdetection

Elephant flow detection in SDN enabled networks

entropy

Traffic analysis through datagram entropy calculation and cluster identification with DBSCAN

f-stack

F-Stack is an open source high performance network framework based on DPDK.

federatedlearning

Thesis in Federated Learning using an Edge/Cloud Computing architecture

filetransfer-using-mpquic-protocol

Setup MP-QUIC structure. Create a topology as below :3. Create the files client-multipath.go and server-multipath.go 4. The files are stored in location “storage-server” and are accessible by server. 5. To execute the code, follow the below steps : Run the python file exported from topology using command : - sudo python2 setup-topology.py Once the terminal enters mininet, open xterm of server and client : - xterm server client In xterm of server, run the below commands : - go build server-multipath.go - ./server-multipath storage-client Here, storage-client is the destination location where client stores its files. In xterm of client, run the below commands : - go build client-multipath.go - ./client-multipath storage-server/abc.txt 100.0.0.1 Here, 100.0.0.1 is the IP address of node - server and abc.txt is the file requested by client which is stored in location “storage-server”...Priority-Based Stream SchedulingIn MP-QUIC, when multiple streams share common path, there are chances of Inter-stream blocking, which may have severe consequences. Also, stream features such as Bandwidth,RTT Delay etc are not taken into consideration when streams are directed on paths in network. In Priority Based Scheduling, instead of making all streams competing for the fast path in a greedy fashion, we allocate paths for each stream by considering the match of stream and path features in the scheduling process. In this, streams can be prioritized by giving them priority value based on path features(bandwidth, RTT, Completion time, etc.) and then the scheduler allocates the new stream to each path with a calculated amount of data. This type of scheduling reduces the burst transmission of packets in congested path or paths with having low delay .Implementation To implement Priority based scheduling, we have assigned a priority to each stream that is created for file transfer / transfer of all the packets created. The streams are scheduled on the basis of decreasing priority of streams on a common path. For example - If 3 streams are created, and we have 2 paths available in network, then, 2 streams will be redirected to 2 paths available, and 3rd stream will be sharing the path with either Stream1 / Stream2. In this case, the stream scheduling is done on the basis of priority. The probability of a stream being selected is calculated by dividing its priority by the priority sum of all the scheduled streams on this path. Network Assisted Path schedulingThe scheduling algorithm we have now, considers the path features, RTT and many things, on the sender side. During the initiation of connection between client and server, the RTT delay is taken and fixed length cells are sent along with handshake. When a destination host receives an RM cell, it will send the RM cell back to the sender with its CI and NI bits intact. With all the information indicated by bit indicators(such as path delay,path congestion,bandwidth etc), packets are scheduled on the path The data cells will be triggered after every constant time so that scheduler can schedule the data in the best path.

flare

An analytical framework for network traffic and behavioral analytics

flow-transmission

:hourglass_flowing_sand:A task scheduler to schedule the transmission of multiple traffic flows

flow_cytometry_analysis_k_means

Analysis and clustering of big data using Hadoop and Flink programming using K-mean on Flow Cytometry Data to identify cells exhibiting similar behaviour and cluster together and remove outliers to identify characteristics of cells which may lead to cancer.

flow_scheduling_dqn_pytorch

Intelligent Load Scheduling Policy for multi-controller sdn

flowatcher-dpdk

In the repository, we present FloWatcher-DPDK, a lightweight software traffic monitor based on Intel(R) DPDK

floweridentification

Neural Network to identify flowers. Based on the PlantCLEF dataset.

flowinspect

A Network Inspection Tool

flowscope

FlowScope is an oscilloscope for your network traffic. It records all traffic continuously in a ring buffer and dumps specific flows to disk on trigger events.

fnss

Fast creation and configuration of topologies, traffic matrices and event schedules for network experiments

fogbus2

FogBus2: A Lightweight and Distributed Container-based Framework for Integration of IoT-enabled Systems with Edge and Cloud Computing

future-internet

Multiple projects about congestion control, adaptive bitrate streaming, topology design, and WAN traffic engineering.

global-optimization-for-optimal-power-flow-over-transmission-networks

gman

GMAN: A Graph Multi-Attention Network for Traffic Prediction

gnn-routing-diss

Dissertation for Computer Science Tripos Part III (MEng) at the University of Cambridge. Associated code is in the gnn-routing repository.

gnnpapers

Must-read papers on graph neural networks (GNN)

gon

[NeurIPS-W'21] Generative Optimization Nets for Memory-Efficient Data Generation

graph-network-properties

Network Properties in Spark GraphFrames In this project, you will implement various network properties using pySpark and GraphFrames. You may need to look into the documentation for GraphFrames and networkx to complete this project. In addition to implementing these network metrics, you will be required to answer some questions when applying these metrics on real world and synthetic networks. Submission Requirements Please create a zip file containing ONLY the following for this project : 0. Do NOT include the data or any extra files. 1. degree.py 2. centrality.py 3. articulations.py 4. A pdf document containing answers to the questions asked in the project description. [Power law questions for Stanford and the 4 random graphs provided in degree.py, answers for the two questions on centrality, answer for the closeness question). Name this as <your_unity_id>.pdf 5. [OPTIONAL] Script for checking power law (if you have written something). Instructions to run in the pdf document which contains the answers to the questions asked in project description. 6. Output for centrality.py saved in a csv file called centrality_out.csv 7. Output for articulations.py saved in a csv file called articulations_out.csv For articulations problem, you may just run it with networkx approach (5 mins run time) and save the output in the file articulations_out.csv. We will be evaluating the project using a script. So please make sure you follow the instructions, or your assignment might not get graded. Degree Distribution The degree distribution is a measure of the frequency of nodes that have a certain degree. Implement a function degreedist, which takes a GraphFrame object as input and computes the degree distribution of the graph. The function should return a DataFrame with two columns: degree and count. For the graphs provided to you, test and report which graphs are scalefree, namely whose degree distribution follows a power law, at least asymptotically. That is, the fraction P(k) of nodes in the network having k connections to other nodes goes for large values of k as P(k) ~ k−γ where γ is a parameter whose value is typically in the range 2 < γ < 3, although occasionally it may lie outside these bounds. Answer the following questions: 1. Generate a few random graphs. You can do this using networkx’s random graph generators. Do the random graphs you tested appear to be scale free? (Include degree distribution with your answer). 2. Do the Stanford graphs provided to you appear to be scale free? Note that GraphFrames represents all graphs as directed, so every edge in the undirected graphs provided or generated must be added twice, once for each direction. When finding if the degrees follow a power law you can then use either the indegree or outdegree, as they should be equal. Centrality Centrality measures are a way to determine nodes that are important based on the structure of the graph. Closeness centrality measures the distance of a node to all other nodes. We will define the closeness centrality as CC(v) =1/ ∑ d(u,v) u∈V where d(u, v) is the shortestpath distance between u and v. Implement the function closeness, which takes a GraphFrame object as input and computes the closeness centrality of every node in the graph. The function should return a DataFrame with two columns: id and closeness. Consider a small network of 10 computers, illustrated below, with nodes representing computers and edges representing direct connections of two machines. If two computers are not connected directly, then the information must flow through other connected machines. Answer the following questions about the graph: 1. Rank the nodes from highest to lowest closeness centrality. 2. Suppose we had some centralized data that would sit on one machine but would be shared with all computers on the network. Which two machines would be the best candidates to hold this data based on other machines having few hops to access this data? Articulation Points Articulation points, or cut vertices, are vertices in the graph that, when removed, create more components than there were originally. For example, in the simple chain 1-2-3, there is a single component. However, if vertex 2 were removed, there would be 2 components. Thus, vertex 2 is an articulation point. Implement the function articulations, which takes a GraphFrame object as input and finds all the articulation points of a graph. The function should return a DataFrame with two columns, id and articulation, where articulation is a 1 if the node is an articulation point, otherwise a 0. Suppose we had the terrorist communication network given in the file 9_11_edgelist.txt . If our goal was to disrupt the flow of communication between different groups, isolating the articulation points would be a good way to do this. Answer the following questions: 1. In this example, which members should have been targeted to best disrupt communication in the organization?

graphml-to-mininet

hedera

Implementing Hedera with Ryu controller.

hefesto

Hefesto es un software de adquisicion de datos para dispositivos edge-computing de alto nivel.