High-availability Seamless Redundancy (HSR)

An HSR node with two ring ports sends an identical frame into both directions simultaneously. When a destination node receives the first frame, it processes it and discards the duplicate arriving from the opposite direction. Non-destination nodes act as bridges, forwarding frames onward. This ensures that if a single link or node fails, the frame always arrives via the alternate path with zero switchover time, a critical requirement for GOOSE trip messages and Sampled Values.

Every HSR frame is appended with a specific HSR tag inserted after the source MAC address. This tag contains:

• Path identifier: Distinguishes between PRP and HSR frames.
• Sequence number: A 16-bit counter incremented for each frame sent by a source.
• Link service data unit size: For backward compatibility.

The destination node uses the source MAC and sequence number pair to identify and discard duplicates, preventing network storms and ensuring only one copy is processed by the application.

A QuadBox is a specialized device used to interconnect multiple HSR rings or to connect an HSR ring to a standard Ethernet network. It functions as a node in two separate rings, selectively forwarding frames between them while respecting the duplicate-discard algorithm. A QuadBox prevents frames from looping endlessly between rings by tracking the source and sequence number of forwarded frames, maintaining the seamless redundancy property across larger, segmented substation networks.

HSR defines specific node roles to integrate diverse equipment:

• DANH (Doubly Attached Node with HSR): A native HSR device with two physical ring ports, such as a modern protection IED or Merging Unit. It generates and processes HSR-tagged frames directly.
• RedBox (Redundancy Box): A converter that allows a standard, single-port Ethernet device (a Singly Attached Node or SAN) to connect to the HSR ring. The RedBox handles all duplication and duplicate-discard logic on behalf of the legacy device, enabling seamless integration of non-HSR-aware equipment.

While both HSR and Parallel Redundancy Protocol (PRP) achieve zero-time recovery through duplication, their topologies differ fundamentally:

• PRP: Uses two completely independent, parallel star-topology networks (LAN A and LAN B). Every node connects to both networks. This requires a duplicated switch infrastructure.
• HSR: Uses a single ring topology without dedicated switches. Nodes are connected directly in a ring, reducing hardware cost and cabling but requiring all nodes to actively forward traffic, which increases the processing load on each device.

HSR nodes continuously monitor ring integrity by sending periodic supervision frames. These frames circulate the ring and allow each node to detect failures such as a link break or a silent node. When a fault is detected, no reconfiguration is needed because the duplicate frame already traversed the alternate path. However, the supervision mechanism provides an alarm to the Substation Automation System (SAS) operator, indicating that the ring is operating in a degraded state and requires maintenance before a second failure could isolate a segment.

A legacy network redundancy protocol (IEEE 802.1w) that dynamically reconfigures Ethernet paths after a link failure. Unlike HSR's zero-time recovery, RSTP requires milliseconds to seconds to converge on a new topology, during which frames are lost. This recovery gap makes RSTP unsuitable for time-critical substation functions like tripping signals, where even a single lost GOOSE message could delay fault clearance.

• Recovery time: Typically 100ms–2s vs. HSR's 0ms
• Mechanism: Blocking redundant paths to prevent loops
• Use case: Non-critical SCADA and engineering access networks

A ring-based redundancy protocol defined in IEC 62439-2 that uses a Media Redundancy Manager (MRM) to monitor ring integrity. Upon detecting a break, the MRM unblocks its secondary port, restoring connectivity within 200–500ms. While faster than RSTP, MRP still incurs a recovery gap, making it less suitable than HSR for substation process bus applications where zero-frame-loss is mandatory.

• Topology: Single ring with MRM node
• Recovery: 200ms typical, configurable down to 30ms
• Comparison: HSR eliminates the MRM single-point-of-failure

Generic Object Oriented Substation Event (GOOSE) messages are the primary payload carried over HSR rings in IEC 61850 substations. These publisher-subscriber frames transmit protection tripping, interlocking, and breaker status signals with strict latency requirements (typically <3ms). HSR's bidirectional frame forwarding ensures that even if the ring breaks at any single point, GOOSE messages reach all subscribers via the alternate path with zero retransmission delay.

• VLAN tagging: Prioritizes GOOSE traffic (IEEE 802.1Q priority 4-7)
• Retransmission: GOOSE repeats at increasing intervals; HSR prevents gaps
• Typical ring: 15–50 IEDs per HSR ring for latency compliance

A QuadBox is a specialized device that interconnects two HSR rings or bridges an HSR ring to a PRP network, enabling scalable redundancy architectures across large substations. It implements the HSR-to-HSR or HSR-to-PRP coupling function defined in IEC 62439-3, forwarding frames between rings while preventing duplicate circulation. This allows substation designers to segment large IED populations into multiple manageable rings without sacrificing seamless redundancy.

• Function: Ring coupling without single-point-of-failure
• Use case: Bay-level rings connected to station-level backbone
• Alternative: RedBox for connecting single-attached SCADA devices