How Cloud Computing Can Support On-Demand Assistive Services
Davide Mulfari, Antonio Celesti, Antonio Puliafito, Massimo Villari · 2013 · Proceedings of the 10th International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/2461121.2461140
Summary
This paper from the University of Messina investigates how cloud computing and virtualisation can provide on-demand assistive technology services for people with disabilities who need to use shared or public computers. The core problem is that when a disabled person uses a computer at an internet café, library, or campus lab, local security policies typically prevent them from installing or configuring the assistive software they need (screen readers, magnifiers, keyboard customisation). Even if they can make changes, all settings must be restored when they leave. The authors propose an open-source architecture that runs personalised assistive software on a remote Virtual Machine (VM) accessible via any standard HTML5 web browser, requiring no local software installation. The architecture uses noVNC (an open-source HTML5 VNC viewer) to display a remote Windows XP desktop running within Oracle VM VirtualBox, accessed through a WebSocket proxy server. The system was tested with three categories of assistive software: keyboard accessibility preferences (MouseKeys for numeric keypad mouse control, FilterKeys for key repeat customisation), the NVDA open-source screen reader, and Microsoft Magnifier screen magnification.
Key findings
The initial testing revealed significant I/O challenges. Keyboard accessibility preferences did not work through VNC because the keyboard events were intercepted locally before reaching the VM, requiring the user to modify local system settings — defeating the purpose. The NVDA screen reader ran on the VM but VNC does not support audio output, so the user could not hear it. Only the screen magnifier worked correctly, with the HTML5 Canvas tag successfully rendering the enlarged desktop within the browser, including colour inversion. To address the keyboard issue, the authors developed a JavaScript-based keyboard accessibility layer within the web application itself, implementing MouseKeys and FilterKeys as browser-side features that translate keyboard events into appropriate VM mouse/keyboard events, with settings stored in personal cloud profiles. For the audio problem, they implemented a streaming solution using PulseAudio, GStreamer, Lame MP3 encoding, and IceCast2 to redirect VM audio output through the browser — though they acknowledge that excessive latency currently makes this unusable in practice. The paper references Cloud4all and GPII (Global Public Inclusive Infrastructure) as related initiatives aiming to provide cloud-based accessibility paradigms.
Relevance
This paper addresses a real and persistent problem: the portability of assistive technology configurations. People with disabilities who rely on customised software setups face a significant barrier when using any computer other than their own — a barrier that sighted, non-disabled users rarely consider. While the specific VNC-based approach described here has been largely superseded by advances in cloud computing (modern Desktop-as-a-Service platforms, Chrome OS with built-in accessibility, and the GPII/Morphic preference portability system), the underlying vision remains relevant: assistive technology should follow the user, not be tied to a specific device. The honest reporting of what didn't work (keyboard preferences, screen reader audio latency) is valuable for the field. The concept of "Assistive Technology as a Service" (ATaaS) — where personalised accessibility configurations are stored in the cloud and delivered on demand to any device — anticipates the direction that projects like Morphic (formerly GPII) have since pursued. For organisations managing shared computing environments (libraries, educational institutions, co-working spaces), ensuring that users can access their assistive technology configurations remains an important accessibility consideration.
Tags: cloud computing · assistive technology · virtual desktop · screen reader · screen magnifier · motor accessibility · HTML5 · remote desktop