Home     Publications

Software available through this page can be freely used and modified. I invite you to notify its use to the authors and acknowledge it in eventual publications (articles, other softwares, presentations).

Dynamic Backup Workers in PyTorch

The software introduces in PyTorch the support for a new dynamic backup worker scheme, where the pa- rameter server decides at each iteration how many stochastic gradients to wait for before updating the parameter vector. This strategy helps to mitigate the straggler problem. Workers can be in the same machine (different cores or GPUs) or distributed across different machines. The code makes also possible to simulate the effect of different distributions for the computation times.
Repository

• Dynamic Backup Workers for Parallel Machine Learning [editor], [preprint]
  Chuan Xu, Giovanni Neglia, and Nicola Sebastianelli, Elsevier Computer Networks, 2021

• Dynamic Backup Workers for Parallel Machine Learning [editor], [preprint]
  Chuan Xu, Giovanni Neglia, and Nicola Sebastianelli, IFIP Networking, June 22-25, 2020.

BT Peers Harvester

Given a torrent file, this software contacts all the trackers to retrieve a specified percentage of the peers of the swarm. It is written in C and the source files are in the src directory. Openssl libs are needed to compile it.
Code. How to compile. How to use.

• Availability in BitTorrent Systems [pdf]
  G. Neglia, G. Reina, H. Zhang, D. Towsley, A. Venkataramani, J. Danaher, 26th Annual IEEE Conference on Computer Communications (INFOCOM) , May 6-12, 2007, Anchorage, Alaska, USA, a longer version is available as UMass Technical Report 06-41 [pdf]
  

Our TCP Marking scheme in NS2

A C++ patch (1000 lines) which implements our TCP marking algorithm (policy WBE2) in the network simulator ns2.1b9a (gcc3.2). Details about the algorithm are in An Analytical Model of a new Packet Marking Algorithm for TCP flows, G. Neglia, G. Bianchi, V. Falletta, Elsevier Computer Networks, Volume 50, Issue 8, 6 June 2006, Pages 1176-1191 [pdf].
Code. Guidelines are in the patch as comments.

• An Analytical Model of a new Packet Marking Algorithm for TCP flows [pdf]
  G. Neglia, V. Falletta, G. Bianchi, Elsevier Computer Networks, Volume 50, Issue 8, 6 June 2006, Pages 1176-1191,special issue on best papers from QoS-IP 2005

• An Analytical Model of a new Packet Marking Algorithm for TCP flows [ps] [ps(rar)]
  G. Neglia, V. Falletta, G. Bianchi, QoS-IP, 2-5 January 2005, Catania, Italy

• An Analytical Model of a new Packet Marking Algorithm for TCP flows: preliminary insights [ps] [ps(rar)] [extended ps] [extended ps(rar)]
  G. Neglia, G. Bianchi, V. Falletta, ISCCSP 2004, 21-24 March 2004, Hammamet, Tunisia

WWW traffic generator for NS2

A tcl script to simulate WWW traffic in NS2. The script creates TCP flows with exponential interarrival times and sizes drawn from an empirical distribution. TCP sources that have completed their transfer are re-used in order to spare memory. For each new source, a new node is created and routing is updated. New sources could also be linked to existing nodes, but we needed to use NS DiffServ markers which act only on a per-IP basis. A more detailed description is at the begin of the script.
Code. Guidelines are in the patch as comments.

• A new Three Color Marker for TCP flows [pdf]
  G. Neglia, G. Bianchi, M. Sottile, Journal of Systemics, Cybernetics and Informatics, Volume 2 - Number 6, 2004

• Performance Evaluation of a new Adaptive Packet Marking Scheme for TCP over DiffServ Networks [pdf]
  G. Neglia, G. Bianchi, M. Sottile, Globecom 2003, Dec. 2003, San Francisco, CA, USA

• A new Three Color Marker for TCP flows [ps] [ps(rar)]
  G. Neglia, G. Bianchi, M. Sottile, CCCT'03, July 31, August 1-2, 2003, Orlando, FL, USA