Example Pipeline

This pipeline details the processing graph configuration for detecting hippocampal population bursts and decoding replay content in real time using the Falcon server framework.

Pipeline Overview

The pipeline processes multi-channel streaming data to detect sharp-wave ripples (SWRs) and trigger closed-loop feedback mechanisms under 1 millisecond of latency.

[OpenEphys/Neuralynx] -> (NlxReader) -> (MultiChannelFilter) -> (RippleDetector) -> (DigitalOutput)

System Requirements

Ensure the server instance allocates dedicated CPU resources to avoid context switching thread overhead:

• Operating System: Linux Enterprise / Ubuntu LTS
• Processor Affinity: Minimum 4 isolated CPU cores
• Data Sources: Open Ephys GUI or Neuralynx hardware stream

---

Configuration Specification

Save the configuration below as hippocampal_replay_pipeline.yaml.

graph:
  name: HippocampalReplayDetection
  sample_rate: 32000
  channels: 128
 
nodes:
  - id: reader_0
    type: NetworkReader
    port: 5555
    data_type: MultiChannel
 
  - id: filter_ripple
    type: BandpassFilter
    low_cutoff: 150
    high_cutoff: 250
    order: 4
 
  - id: detector_swr
    type: RippleDetector
    threshold_sd: 5.0
    min_duration_ms: 15
 
  - id: sink_ttl
    type: DigitalOutputSink
    hardware_line: TTL_0
 
connections:
  - from: reader_0:output
    to: filter_ripple:input
  - from: filter_ripple:output
    to: detector_swr:input
  - from: detector_swr:event
    to: sink_ttl:input

---

Processing Nodes Reference

NetworkReader

Initializes socket streams over TCP/IP protocol layers to receive raw binary matrices from recording amplifiers.

BandpassFilter

Applies a zero-phase forward Butterworth filter across the assigned MultiChannel data types.

RippleDetector

Tracks individual signal envelopes using root-mean-square transformations. It continuously updates the moving baseline variance metrics.

---

Validation and Verification

Run the validation script to verify the graph loop optimization topology before production deployment:

falcon-cli --validate -f hippocampal_replay_pipeline.yaml
falcon-server --config hippocampal_replay_pipeline.yaml --daemon

Expected Metric Layout

Metric	Lower Bound	Target Range	Upper Bound
Latency	0.2 ms	0.4 ms	0.8 ms
CPU Core Jitter	1 μs	3 μs	10 μs
Memory Overhead	45 MB	64 MB	128 MB

Troubleshooting Matrix

1. Buffer Overflows: Increase the thread priority or isolate the dedicated hardware cores using boot arguments.
2. Dropped Network Packets: Verify system network buffer allocations using kernel performance profiles.