EVS

 

EVS processes a camera feed, it doesn't just throw an alert; it translates visual actions into semantic data triples (e.g., [Vehicle X] ---> [Loitered Near] ---> [Gate 4]).

Because it runs on the same high-throughput P11 compute node, these triples are immediately injected into the KGNN database. If a suspicious vehicle drops off a package, EVS flags the visual, and KGNN instantly cross-references it with your logistics logs, badge-reader databases, or local intelligence manifests in real-time.

The Bottom Line

Instead of relying on Dahua or Hikvision to provide closed-ecosystem, high-risk "smart" NVR software, Equitus EVS on IBM P11 shifts the intelligence upward. It allows global operators to leverage cheaper, ubiquitous Asian camera hardware for basic capture, while the actual heavy-duty video intelligence, data compliance, and security profiling remain completely privatized, local, and bulletproof.

• Dual-Homed Network Topologies: The IBM Power11 hardware running EVS is deployed with multiple physical network interfaces (NICs). One interface sits strictly inside a dead, non-routable CCTV subnet with the NVRs, blocking any internet access. The other interface faces upward into the secure enterprise network where the KGNN platform resides.

• Proxy Re-streaming: EVS acts as an internal, sanitized proxy. It consumes the raw, high-bandwidth streams locally, strips out the risk, and forwards only structural metadata (or highly secure, transcoded MJPEG/H.264 clips) to the rest of the enterprise stack.

3. Deep-Storage "Cold" Forensics via Virtual VMS Integrations

If the goal is to query recorded archive video rather than a live stream, EVS bypasses the NVR's custom local search apps by interfacing directly with storage:

• Network File System (NFS) / SMB Scraping: EVS can mount the underlying storage directories of the NVRs if they are exposed via network shares. It parses raw video files directly from the storage disks, skipping the NVR's software layer entirely.

• VMS Universal Drivers: EVS utilizes open Video Management Software (VMS) drivers to act as a universal translation bridge. It treats the non-USA NVR as a generic storage array, standardizing the file-retrieval pipeline.

4. Direct Camera Interception (Bypassing the NVR Entirely)

In scenarios where the non-USA NVR's onboard firmware is entirely locked down, corrupted, or deemed too high-risk to communicate with, EVS can bypass the recorder altogether:

• Network Tapping / Switch Mirroring: Utilizing a managed switch, network engineers mirror the traffic from the physical camera ports directly to the EVS ingestion node. EVS steals a split of the raw camera feeds via unicast or multicast, letting the non-USA NVR continue recording to its local hard drives in an isolated silo while EVS performs real-time threat detection on the mirrored data.

The Outcome for the KGNN

By standardizing the input at the EVS layer, it doesn't matter if the raw video originated from an NDAA-compliant US camera or a 10-year-old foreign NVR. EVS sanitizes the input stream, maps the pixels to local vector transformations inside the Power11 matrix multiply engine, and outputs identical, sanitized Structured Triples to the Equitus KGNN.

The Knowledge Graph remains 100% insulated from the underlying physical hardware, seeing only clean entity nodes like (Camera_Zone_B) --[Detects]---> (Unscheduled_Vehicle).