Chapter 1. System Components

Architecture, component functions, working principles, and deployment boundaries

1.1 System Architecture

The road video surveillance system is organized around four deployment boundaries: the Field Side, the Edge Side, the Central Side, and the External Side. This boundary model is fundamental to understanding responsibility, failure isolation, and interface design. The Field Side encompasses all physical sensing and mounting hardware deployed at road points. The Edge Side aggregates field devices at roadside cabinets and provides local protection and connectivity. The Central Side hosts all platform services, storage, and analytics. The External Side represents partner systems that exchange data with the surveillance platform through defined interfaces.

Deployment boundary discipline is critical: the system boundary includes all devices from camera lens to storage media and platform services. External systems are integrated via defined interfaces and must not be tightly coupled to avoid cascading failures. This means that a failure in the toll system, for example, must not prevent the surveillance platform from recording or exporting evidence independently.

Figure 1.1: Four-zone system architecture — Field Side, Edge Side, Central Side, and External Side with data flow arrows and core/optional component markings

Core Required vs. Optional: Camera, edge switch, central VMS, storage cluster, and time synchronization are mandatory in every deployment. Edge NVR, thermal cameras, radar triggers, and 5G backup are optional and should be selected based on scenario requirements and budget constraints.

1.2 Component List & Functions

The system comprises seven major subsystem categories, each with distinct inputs, outputs, key performance indicators, and mismatch risks. The component breakdown tree below illustrates how each category decomposes into specific devices and their associated performance parameters. Understanding these KPIs and mismatch risks is essential for procurement, acceptance testing, and troubleshooting.

Figure 1.2: Component breakdown tree — Road Surveillance System decomposed into seven subsystem categories with KPIs and sub-components

Key Component Specifications

The following table provides a structured overview of each major component category, covering its primary I/O interfaces, key performance indicators, and the consequences of specification mismatches. This table should be used as a checklist during design review and procurement.

Component	Primary I/O	Key Performance Indicators	Mismatch Risk
ANPR / Capture Camera	Trigger input (radar/virtual line); snapshot + metadata output; dual video streams	Plate capture rate ≥95%; shutter ≤1/1000s; WDR ≥120dB; IR performance at speed	Wrong focal length → insufficient pixels on plate; poor IR → overexposure at night
Fixed Bullet Camera	RTSP/ONVIF stream; alarm inputs (optional); device telemetry	WDR ≥120dB; low-light SNR ≥52dB; IP66/67; operating temp -30°C to +60°C	Consumer-grade outdoor use → fogging, poor reliability, early failure
PTZ Camera	Video stream; PTZ control; preset/tour commands	Rotation speed ≥200°/s; preset accuracy ±0.1°; optical zoom ≥20×; stabilization	Insufficient stabilization on bridge vibration → unusable zoom images
Illuminators (IR/White)	Power input; photocell/remote control	Illumination uniformity ≥80%; wavelength match to sensor; thermal management	Glare to drivers; plate overexposure; regulatory non-compliance
Roadside Cabinet / Pole Box	Power distribution; fiber termination; patch panel	IP55 min (IP66 preferred); corrosion resistance; thermal design ≤55°C internal	Undersized cabinet → overheating, maintenance impossible, equipment failure
Industrial PoE Switch	PoE ports (8–24); SFP uplinks; VLAN/QoS; ring protocols (ERPS/RSTP)	PoE budget ≥240W; surge immunity ≥4kV; -40°C to +75°C; ring convergence <50ms	Insufficient PoE budget → camera brownout; no ring support → large outage on fiber cut
Backhaul (Fiber/Wireless/Cellular)	Ethernet/IP transport; primary and backup paths	Throughput ≥1G (10G for dense nodes); latency <5ms; availability ≥99.9%; protection switching <50ms	Oversubscription → dropped frames, export failure, evidence gaps
Central Storage (NVR Cluster / VMS Storage)	Ingest streams; store; playback/export	Write bandwidth ≥1.2× peak ingest; retention 30–90 days; RAID-6 or erasure coding; rebuild time <24h	RAID rebuild too slow → risk of second disk failure causing data loss
VMS Platform	Device management; user access; events; REST/SDK APIs	Concurrent views ≥100; search speed <3s; alarm throughput ≥1000/min; audit log integrity	Licensing model mismatch → cost explosion at scale; API incompatibility → integration failure
Time Sync (NTP/PTP)	Time signal distribution; drift monitoring	Drift ≤1s across all devices; stratum hierarchy; holdover ≥24h on GPS loss	Time inconsistency → evidence disputes, ANPR correlation failures, audit gaps

1.3 Working Principle

Startup Sequence

On power-on, the roadside cabinet initiates a controlled startup sequence. The switch boots first and begins PoE sequencing to cameras, which prevents simultaneous inrush current from tripping breakers. Cameras perform a power-on self-test, acquire their IP address via DHCP or static configuration, and register with the VMS. The VMS validates device identity, pushes configuration templates (stream profiles, recording policies, analytics zones), and begins ingesting streams. Time synchronization is established within the first minute; devices that fail to sync within the configured threshold are flagged for operator attention.

Normal Operation

During normal operation, cameras stream continuously to the edge switch, which forwards traffic over the backhaul ring to the central ingest service. The storage cluster writes continuous recording streams and event clips simultaneously. The analytics engine processes video streams — either at the edge or centrally — and produces alarm events with metadata. The VMS correlates alarms with GIS coordinates and presents them to operators on the monitoring wall. Operators can view live feeds, review recordings, export evidence packages, and issue PTZ control commands. All actions are logged in the audit trail.

Exception Handling

The system is designed to handle three major exception classes gracefully, each with defined detection, response, and recovery procedures.

Exception	Detection	Immediate Response	Recovery	Acceptance Test
Backhaul Fiber Cut	VMS ingest alarms; packet loss spike on ring segment	Ring protocol reroutes traffic in <50ms; if edge NVR present, local caching begins	After link restoration, backlog sync uploads to central; evidence gap occurs if no edge storage	Alarm within defined seconds; operator notification; no config loss
Lightning Surge on Pole	SPD clamps surge; PoE switch port may trip; camera offline alarm	UPS ride-through maintains power; auto-reclose breaker; switch PoE port power-cycle	Repeated surge triggers require grounding resistance re-test and SPD replacement	Camera recovery time; SPD indicator check; grounding resistance measurement
Time Server Drift/Failure	VMS detects drift > threshold; devices hold last sync	VMS blocks evidence export or annotates "time untrusted"; operator alert	Switch to backup NTP; correct drift; re-sign evidence chain	Drift log review; export block verification; backup NTP failover test

System Boundary Note: The surveillance system explicitly does not replace traffic engineering devices (signal controllers, lane controllers, barrier gates). Integration with these systems must use defined interfaces with retry and fallback logic. A failure in an external system must never cause the surveillance system to stop recording or lose evidence.