CCGSC

Crowd Cloud Grid Super-Computing System (CCGSC)

A hierarchical, decentralized platform that pools idle computing power from everyday devices, web browsers, and servers to make high-performance computing more affordable and accessible.

What it is

The Crowd Cloud Grid Super-Computing System (CCGSC) is a distributed computing architecture designed to democratize high-performance computing (HPC). It connects virtual machines (VMs), containers, sandboxes, and browser-based worker nodes across Dew, Cluster, Grid, and Cloud/Web layers so heterogeneous devices—from mobile phones and personal computers to servers—can contribute idle CPU, GPU, RAM, and storage resources while still serving everyday personal or business use.

Why

Problem: Traditional supercomputers and centralized cloud platforms carry high capital, operational, and energy costs, while existing volunteer computing networks leave large amounts of global idle computing capacity underutilized and struggle with variable, intermittent nodes.

Goal: Design and evaluate a decentralized platform for affordable, resilient high-performance computing that supports secure resource sharing on heterogeneous devices, offline capability, and massively parallel or probabilistic workloads (such as Monte Carlo simulations and scientific computing).

Central research question: How does CCGSC's P2P/blockchain hierarchy improve utilization over traditional systems?

How it works

  • Four-layer hierarchy: Organizes participating devices across a Dew layer (low-power or intermittent devices using sandboxes and offline processing), a Cluster layer (dynamic VM orchestration for compute-intensive work), a Grid layer (high-availability servers), and a Cloud/Web layer (global coordination).
  • Dynamic resource pooling: Aggregates available CPU, GPU, RAM, and storage through coordinating head nodes using regular heartbeat signals and first-in, first-out (FIFO) queues.
  • Independent parallel workloads: Splits large probabilistic and embarrassingly parallel workloads into smaller, independent subtasks that run concurrently across distributed nodes and synchronize results when online.
  • Isolated VMs and zero-install browser nodes: Supports full-OS virtual machines and containers alongside web browser worker nodes powered by WebAssembly and Web Workers, allowing participants to contribute compute capacity without disrupting normal device use.
  • Peer-to-peer distribution and lightweight verification: Uses peer-to-peer (P2P) protocols for decentralized task and data sharing, combined with lightweight blockchain consensus for trustless authentication.

Status

Research prototype, work in progress. The system architecture and prototype components use containerization (Docker/Podman) and browser-based execution (WebAssembly) to evaluate task distribution, throughput, and latency across local and wide-area networks.

Live demo

Explore the interactive demonstration environment:

https://ccgsc-demo.digitalbd.org/

Author

Shyed Shahriar

Contact

https://digitalbd.org/