Alex JF


  • Category: Distributed Systems
  • Authors: Alexandre Fonseca, Pedro Luz, João Quitério, David Forte, Hugo Almeida, Rui Silva
  • Start date: 2012/02
  • Version: 1.0
  • Status: Finished

This project was created as an assigment for the Software Engineering and Distributed Systems course during my 6th semester at IST.

In this project we were tasked with the creation of a mobile communications network manager composed of several servers, tolerating one server failure (crash or byzantine behaviour) even in the presence of big network delays.

The project was split in 4 parts:

  • Domain layer design, implementation and testing: Cellphone, Network, Communication, etc.
  • Service layer design, implementation and testing: SendSMS, SendCall, etc.
  • Development of a web interface for the system using GWT and implementation of network servers using JAX-WS under JBoss 7 found through a UDDI server. Service interface through SOAP.
  • Implementation of non-functional requirements for the distributed servers: tolerate 1 byzantine or non-byzantine failure and provide adequate security. This last part was managed through the Scrum methodology.

Failure tolerance was implemented through a quorum-based active replication protocol employing a quorum of 4 nodes among the 5 replicas.

The security requirements were that attempts to tamper or replay messages had to fail to compromise the system. To that end, all messages exchanged between the server cluster are digitally signed and certified by a Certification Authority system which we also had to implement. The prevention of replay attacks was accomplished by introducing a UUID unique to each message and duplicate detection.


comments powered by Disqus