This visualizer simulates a grid of sensor nodes — like the readout network of a particle detector — routing data hop-by-hop toward “root” nodes at the bottom of the grid, which stand in for the readout aggregators. Sensors can fail, links can get congested, and the simulation lets you compare different routing strategies and network layouts under those conditions.

What you’re looking at

  • Each square is a sensor node. Its color is a heatmap of how much data has passed through it (blue = idle, red = heavily loaded); a bar at the bottom of the square fills up as its buffer fills. Purple squares are root nodes.
  • Grayed-out squares with an X are failed nodes — either permanently broken (“static” failures, shown in red) or temporarily glitching (“temporary” failures, which show a countdown until they recover).
  • Thin purple lines are the physical links between neighboring nodes; an orange arrowhead marks the direction data is actively being routed along that link.
  • Click any node to select it — a panel on the right shows its queue, drop count, and the packets it’s currently holding, and traces each packet’s hop-by-hop path across the grid in a matching color. Press F to toggle a static failure on the selected node.
  • The right-hand panel also has Stats (score, throughput, delivered/lost packets, a running score graph, and a hop-count histogram) and Hit Map (for scenarios using real detector data, a heatmap of where hits occurred) tabs.

Running the simulation

  • STEP advances the simulation by one round. Run starts continuous stepping; Run Nx runs a fixed batch of rounds (adjustable with the +/- buttons) and then stops.

Settings you can change

  • Generator — how data gets injected into the network: Random (each node has a chance per round to generate a packet), RngPix (models random pixel hits at a configurable rate), or Root (replays real hit patterns from a detector data file).
  • Router Type — the routing strategy under test: Simple (greedy shortest-path), Smart (congestion-aware routing that avoids overloaded neighbors), Random, and Token (a token-passing scheme).
  • Topology — the shape of the network graph connecting nodes: 2D mesh, diagonal, a mixed diagonal-mesh, hexagonal, octagonal, or the RD53 detector-chip layout.
  • Grid size, buffer size, and failure rates (static and temporary) let you scale the network and dial in how harsh the environment is.
  • EOCN enables a simplified model of end-of-column readout logic downstream of the root nodes, shown as an extra row of buffers below the main grid.

Click Apply & Restart after changing settings to rebuild the simulation with the new configuration.