FADA

Submitted By Fionnuala O\'Donnell on 16 May, 2006

Case Study: FADA (Framework for Animations of Distributed Algorithms)

FADA [1] arose out of need to combat the inadequacies posed by existing teaching methodologies in capturing the dynamic and concurrent behaviour of a distributed algorithm. Its initial aim was to provide a toolkit which instructors could use to not only build and visualise their own representations of an algorithm’s behaviour, but also, to enable them to experiment with its state and behaviour in real time. Such a design arose as a consequence of observing the teaching behaviours of two or more computer science lecturers and the difficulties they encountered in teaching the material.

Based on these observations, the initial design was story boarded and mocked up. This consisted of a framework or library of methods which a user could quickly customise and invoke to reflect the behaviour of the message passing distributed algorithm that he was implementing. This was automatically designed to be outputted as a simulation. However, on presentation to and experimentation by one of the lecturers observed, the toolkit design was amended to offset the code dependency in creating the algorithm definition through the use of a wizard and the provision of a development environment as typified by modern programming environments. The wizard was designed to provide a set of menu items that a user could quickly select and customise to create the messages that a node is to receive, the content of those messages and the type of variables or state information that a node is to store.  The development environment was designed to include an editor that captures on compilation logical and syntactical errors in a user’s algorithm definition code.

Figure 1: A graphical depiction of the manner by which a user deploys the wizard to create a message called ‘Answer’ and to associate with that message a parameter called ‘reply’.

Figure 2: A depiction of the code automatically generated by the wizard in Figure 1. This code is outputted to the code frame within FADA.

Figure 3: A graphical depiction of the manner by which FADA automatically alerts users to syntax errors in their algorithm implementation code. In the above example, the user has forgotten to close the statement in line 14 with a semi colon.

As a consequence of its design, FADA enables users to build dual representations of an algorithm’s behaviour and to map logical errors in the visualisation of an algorithm’s behaviour to that of its implementation or vice versa. It also enables learners to create their own understanding of an algorithm’s behaviour and to experiment with that understanding in real time, learning as they go, in a process known as constructionism. Hence, FADA can be attributed to two modes of use. It can be used by instructors as an interactive presentation tool and it can be used by learners to enable them to create their own simulations of an algorithm’s behaviour.

Figure 4: Screenshot of the dual representations of an algorithm's behaviour afforded by FADA. To the right hand side is the  development environment (Code Frame) where one creates his own algorithm definition. To the left hand side is the simulation automatically rendered by FADA of the algorithm definition. This is held inside what is known as the animation frame.

[1] Fionnuala O'Donnell, "Simulation Framework for the Teaching and Learning of Distributed Algorithms", Department of Computer Science, Trinity College Dublin, PhD, 2006, p.1-260.