MTP3 Congestion Control Through ONPC and Transfer Controlled Messages
Australia's telecommunications landscape stretches across vast distances, from the dense fibre corridors of Sydney and Melbourne to remote microwave links servicing mines in the Pilbara. This geography places unique pressure on signalling infrastructure, where a single congestion event in a major city like Brisbane or Perth can ripple through interconnects carrying voice and data between carriers such as Telstra, Optus and TPG.
Within the Signalling System No. 7 stack, Message Transfer Part Level 3 handles the routing and congestion management that keeps these links healthy. When traffic exceeds capacity, MTP3 activates procedures designed to shed load before packet loss cascades into dropped calls and failed transactions.
Two mechanisms sit at the heart of this protection: the ONPC procedure and the Transfer Controlled message family. Understanding how they interact is essential for engineers maintaining legacy PSTN interconnects, designing SIGTRAN replacements, or auditing signalling security in modern IP-based cores.
Congestion in MTP3 Networks
MTP3 monitors every active signalling link and link set, tracking the occupancy of retransmission buffers and the rate of failed transmissions. When buffer fill crosses configured thresholds, the layer declares congestion at three escalating levels.
These levels are not arbitrary. They map directly to how aggressively the sending side must slow down, and how the receiving peer should respond. The Australian Communications and Media Authority has long required carriers to document these thresholds as part of their network integrity plans under the Telecommunications Act 1997.
The Role of ONPC
ONPC, or Only No Package Congestion, is invoked when a signalling point determines that its transmit buffers are full but no new message packages can be assembled. In practical terms, the node stops originating new SCCP or ISUP user messages while still permitting management and link-state traffic to flow.
This selective freeze protects routing updates and keep-alives, which are critical when managing linksets that span from Adelaide to Darwin across the NBN's transit infrastructure. Without ONPC, a flood of user traffic could starve the very signalling that keeps the network aware of its own topology.
Transfer Controlled Messages Explained
Transfer Controlled, abbreviated TFC, is the message MTP3 uses to tell an adjacent signalling point to throttle a specific destination. The message carries the affected point code and a congestion level indicator, allowing the originator to choose where to apply backpressure.
A TFC can be broadcast to all reachable points or targeted to a single destination that is causing trouble. This targeted behaviour is useful in mixed-vendor environments where an international gateway in Mascot might need relief without affecting domestic routing through a Melbourne-based STP. When load is balanced across a linkset, load sharing across links determines which physical circuit carries the next message, which influences how quickly congestion indicators propagate.
Comparing Congestion Levels and Responses
| Level | Local Action | Peer Notification | Traffic Affected |
|---|---|---|---|
| 0 | Normal operation | None | All flows |
| 1 | Slight throttling | Optional TFC | Non-priority |
| 2 | Moderate throttling | TFC recommended | All user traffic |
| 3 | Full stop | Mandatory TFC | User messages only |
This captures the relationship between declared congestion level, the local throttling behaviour MTP3 enforces, and the messaging obligation toward adjacent signalling points. Level 0 is the steady state and generates no notifications, while Level 3 represents a full local stop with mandatory TFC emission to inform upstream nodes.
Operators must also map these levels to their internal monitoring dashboards so that NOC staff in Sydney or Perth can correlate customer-reported issues with the underlying signalling state. Without this mapping, congestion events may be attributed to endpoint faults rather than link-layer backpressure.
Detection and Threshold Configuration
Detection relies on two counters per link: the number of messages in the transmit buffer and the number of retransmissions required. Once either counter crosses its first threshold, a Congestion Indication primitive is sent upward to MTP3 users.
Carriers operating under the ACMA's Telecommunications (Network Reliability) determinations must configure these thresholds to match documented busy-hour traffic profiles. Sydney's CBD exchanges, for example, see predictable spikes during morning commute hours, which informs buffer sizing decisions.
Signalling Congestion and Recovery
When local congestion is detected, the node generates an internal indication and may emit a TFC toward the source of problematic traffic. The receiving node then applies backpressure by reducing its outbound rate toward the congested destination, preventing the buffers from overflowing entirely.
When buffer occupancy drops below the abatement threshold for a sustained period, MTP3 declares the link de-congested. It then sends a Transfer Controlled with a zero congestion level, or simply stops sending TFCs, allowing the peer to resume normal rates. Recovery must be gradual to avoid oscillation. Australian operators typically tune abatement thresholds lower than onset thresholds to provide hysteresis, a practice that aligns with resilience requirements published for critical infrastructure sectors.
Tuning Practices in Australian Networks
In a market dominated by a few large carriers, consistent tuning practices matter. Engineers at exchanges from Hobart to Cairns rely on shared baselines for buffer thresholds, TFC generation intervals, and ONPC invocation rules.
Local legislation encourages transparency: under the Telecommunications Act, providers must demonstrate that their signalling networks meet reliability obligations, which includes evidence of proper congestion management. Field teams often validate configurations against reference designs before cutover.
Audit the current linkset configuration against the documented thresholds before the next maintenance window, capturing baseline buffer occupancy and TFC emission rates for later comparison.
