Before a single word travels across a circuit-switched network, an elaborate dance must occur. This dance—connection establishment—is the process of creating the dedicated path we examined in the previous page. It involves signaling between endpoints and the network, negotiation of resources, path selection through potentially complex network topologies, and coordination among multiple switches.

Connection establishment is what distinguishes circuit switching from other paradigms. In packet switching, data can flow immediately—each packet carries its own addressing and routing information. In circuit switching, the network must first prepare itself to carry your communication: allocating resources, configuring switches, and verifying end-to-end connectivity.

This page examines connection establishment in meticulous detail, from the moment you lift a telephone handset to the instant you hear the other party say 'Hello.'

Every circuit-switched communication follows a three-phase lifecycle. Understanding these phases is fundamental to grasping how circuit switching operates and where its overhead costs arise.

Phase 1: Circuit Establishment (Call Setup)

Before any user data can flow, the circuit must be established. This phase involves:

• Signaling initiation — The calling party generates a request to establish communication
• Destination identification — The network determines where the call should be routed
• Path selection — Intermediate switches choose and allocate resources for the path
• Resource reservation — Each switch along the path reserves capacity for this connection
• End-to-end verification — The network confirms that a complete path exists to the destination
• Destination alerting — The called party is notified of the incoming call
• Call acceptance — The called party accepts, establishing the bidirectional circuit

Phase 2: Data Transfer

Once the circuit is established, the dedicated path becomes available for communication:

• Transparent transmission — Data flows directly between endpoints without network processing
• Full-duplex operation — Both parties can transmit simultaneously
• Consistent quality — Performance remains stable throughout (no congestion-based degradation)
• No per-unit routing — Unlike packets, each signal segment doesn't need destination information
• Duration varies — Could be seconds (short call) to hours (extended data transfer)

Phase 3: Circuit Disconnect (Teardown)

When communication ends, resources must be released:

• Disconnect initiation — Either party signals the end of communication
• Resource deallocation — Each switch releases its internal connections and trunk allocations
• Signaling propagation — Disconnect signals traverse the path to notify all involved nodes
• Accounting completion — Call duration and resources used are logged for billing
• Resources returned to pool — Capacity becomes available for new connections

Signaling is the mechanism by which endpoints and the network exchange control information to establish, manage, and terminate circuits. It's the 'control plane' that orchestrates the 'data plane' (the actual voice or data transmission).

Types of signaling:

In-band vs. Out-of-band signaling:

A critical architectural decision in signaling design is whether control information shares the same channel as user data (in-band) or uses a separate channel (out-of-band).

In-band signaling:

Historically, telephone networks used in-band signaling, where control tones traveled over the same voice circuit:

• Pulse dialing — Opening and closing the phone line to generate digit pulses
• DTMF tones — Dual-tone multi-frequency signals on the voice path
• Multi-frequency (MF) signaling — Inter-switch signaling using audio tones on trunk lines

Advantages:

• Simple implementation—no separate signaling infrastructure needed
• Works with any transmission medium capable of carrying voice

Disadvantages:

• Limits information capacity (must fit in voice bandwidth)
• Slower (tones must be long enough to detect reliably)
• Security vulnerabilities ('blue box' fraud exploited in-band trunk signaling)

Out-of-band signaling:

Modern networks use out-of-band signaling, where control messages travel on separate channels:

• ISDN D-channel — A 16 or 64 kbps channel dedicated to signaling, separate from voice B-channels
• SS7 signaling links — A completely separate packet-switched network for inter-switch signaling
• SIP messaging — Separate transport connections for session control

Advantages:

• Much faster signaling (full digital speed, not limited to audio)
• Richer information (complex messages, not just tones)
• Secure (users cannot inject control signals into voice path)
• Enables enhanced services (caller ID, call forwarding, etc.)

Disadvantages:

• More complex infrastructure (separate signaling network required)
• Higher capital cost (duplicate network elements)

Signaling System 7 (SS7) is the global standard protocol suite that orchestrates circuit-switched telecommunications. Understanding SS7 reveals how connection establishment actually works in telephone networks at a technical level.

SS7 architecture:

SS7 creates a separate packet-switched network dedicated to carrying signaling messages. This network is distinct from the circuit-switched voice network:

• Signaling Points (SP) — Nodes that send and receive signaling messages
• Signaling Transfer Points (STP) — Packet switches that route signaling messages
• Signaling Links — 56/64 kbps data links connecting signaling points
• Signaling Routes — Paths through the signaling network between endpoints

SS7 protocol layers:

SS7 is organized into layers similar to the OSI model:

Physical Layer: MTP Level 1

• Defines the electrical characteristics of signaling links
• Typically 56 kbps or 64 kbps digital trunk channels

Data Link Layer: MTP Level 2

• Ensures reliable message delivery over individual links
• Error detection, correction, and flow control
• Sequence numbering and acknowledgments

Network Layer: MTP Level 3

• Routes messages through the signaling network
• Handles network congestion and failures
• Provides addressing (signaling point codes)

User Part Layer: ISUP (ISDN User Part)

• Carries call control messages for circuit-switched connections
• Defines messages for call setup, answer, release, and more

Let's trace through a complete call setup, examining what happens at each stage from the moment you lift the handset to when you hear 'Hello' on the other end.

Stage 1: Off-Hook Detection

When you lift the telephone handset, you close a switch that allows current to flow through the local loop (the copper pair connecting your phone to the telephone exchange). The exchange detects this current flow within milliseconds.

Technical detail: The local loop typically carries 20-50 mA of DC current when off-hook (phone in use) versus near-zero when on-hook (phone idle). Line cards at the exchange monitor this current continuously.

Time budget for call setup:

Each stage consumes time:

Post-Dialing Delay (PDD):

The time from completing dialing to hearing ringback is called Post-Dialing Delay. International calls historically had PDD of 5-15 seconds; modern networks target under 3 seconds.

Optimizations:

• Overlap outpulsing — Begin routing before all digits received (for predictable prefixes)
• Trunk forecasting — Pre-select trunks before routing computation completes
• Local exchange bypass — For high-volume routes, use direct trunks between exchanges

In complex networks with multiple paths between any two points, how does the network choose which route a circuit should take? Path selection in circuit-switched networks differs fundamentally from packet routing.

Hierarchical routing in PSTN:

Traditional telephone networks used a hierarchical structure with multiple levels of switches:

Class 5: Local Exchange (End Office)

• Connects directly to subscribers
• Handles local calls internally
• Hands off non-local calls to higher hierarchies

Class 4: Toll Center / Tandem

• Aggregates traffic from multiple Class 5 offices
• Provides routing between local access and long-distance

Class 3/2/1: Primary/Sectional/Regional Centers

• Higher levels in the long-distance hierarchy
• Ensure connectivity between distant regions

Routing strategies:

1. Final Route (Hierarchical):

Every switch has a guaranteed 'final route' following the hierarchy. A call always has a path by climbing up the hierarchy and descending to the destination. This ensures universal connectivity but may use more trunks than necessary.

2. High-Usage Routes:

For common traffic patterns, direct trunks ('high-usage routes') connect switches that frequently exchange calls, bypassing the hierarchy:

• Lower trunk count (more efficient)
• Lower delay (fewer hops)
• Load-balanced with hierarchical routes

3. Dynamic Routing:

Modern networks use dynamic algorithms that consider real-time network conditions:

• Real-Time Network Routing (RTNR) — AT&T's system that queries central databases for optimal routes
• Dynamically Controlled Routing (DCR) — Adjusts trunk selection based on current loads

Trunk hunting sequences:

When selecting an outgoing trunk, switches use hunting sequences that define the order to try trunk groups:

1. Try high-usage route to destination tandem (if available)
2. If blocked, try high-usage route to alternate tandem
3. If blocked, use final route up the hierarchy
4. If all routes blocked, return busy signal

This provides efficiency (use direct routes when available) while ensuring reliability (always have fallback paths).

A fundamental characteristic of circuit switching is that it can block new connection requests. If all resources along any necessary path are in use, the network cannot establish a new circuit—even if there's capacity elsewhere in the network.

Why blocking occurs:

Circuit switching guarantees resources once a circuit is established. This means resources cannot be over-subscribed or borrowed from active calls. When demand exceeds capacity, new requests must be refused.

Blocking probability:

Network engineers design circuit-switched networks to achieve target blocking probabilities:

• Good grade of service: 0.1% - 1% blocking (1-10 in 1000 calls blocked)
• Acceptable for business: 1% - 2% blocking
• Marginal: 2% - 5% blocking
• Unacceptable: >5% blocking

The Erlang B Formula:

Traffic engineering for circuit-switched networks relies on the Erlang B formula, which calculates blocking probability given traffic load and number of circuits:

$$B = \frac{\frac{A^N}{N!}}{\sum_{k=0}^{N}\frac{A^k}{k!}}$$

Where:

• B = Blocking probability
• A = Offered traffic in Erlangs (hours of usage per hour)
• N = Number of circuits (trunks) available

Traffic measurement: Erlangs

The Erlang is the fundamental unit of telephone traffic, representing one hour of usage per hour:

• 1 call lasting 1 hour = 1 Erlang
• 2 calls averaging 30 minutes = 1 Erlang
• 60 calls averaging 1 minute each = 1 Erlang

Example calculation:

If a trunk group carries 10 Erlangs of traffic (e.g., 120 calls/hour averaging 5 minutes each), and you want P(blocking) < 1%:

You need 20 trunks to achieve <1% blocking with 10 Erlangs of traffic.

Call Admission Control (CAC):

Modern circuit-switching-like systems implement explicit Call Admission Control to manage blocking:

1. Pre-admission checking:

• Before attempting circuit setup, verify resources are likely available
• Reject immediately if predicting high blocking probability

2. Congestion throttling:

• When blocking exceeds thresholds, reduce incoming call attempt rate
• Send 'network busy' signals earlier in the call flow

3. Priority reservations:

• Reserve portion of capacity for high-priority traffic (emergency calls, premium customers)
• Lower-priority calls blocked before reserved capacity exhausted

4. Alternate routing with fallback:

• Try multiple routes before declaring blocking
• Different blocking thresholds trigger escalation to alternate routes

These mechanisms ensure that when blocking occurs, it's handled gracefully and recovery is rapid when capacity becomes available.

Connection establishment can fail for many reasons beyond simple blocking. Understanding failure modes and recovery mechanisms is essential for appreciating the robustness engineered into circuit-switched networks.

Types of connection failures:

Failure detection timing:

The speed of failure detection directly impacts user experience. Circuit-switched networks implement multiple timeout levels:

Short timeouts (100-500ms):

• Signaling link timeout — message not acknowledged
• Trunk seizure timeout — no acknowledgment of trunk capture

Medium timeouts (3-10 seconds):

• Address complete timeout — waiting for ACM after sending IAM
• Alerting timeout — waiting for answer after ringing starts

Long timeouts (30-120 seconds):

• No-answer timeout — called party never picks up
• Abandoned call timeout — calling party hangs up during setup

Recovery mechanisms:

1. Automatic Alternate Routing (AAR): When a route fails, automatically attempt pre-configured alternate routes:

Primary: Direct trunk to destination — FAIL
Alternate 1: Via regional tandem — FAIL  
Alternate 2: Via interexchange carrier — SUCCESS

2. Retry with backoff: For transient failures, retry the same route after a delay:

• First retry: 100ms
• Second retry: 500ms
• Third retry: 2 seconds
• Then escalate to alternate route

3. Glare resolution: Glare occurs when both ends of a trunk try to seize it simultaneously for different calls. Resolution:

• Compare switch identities (higher ID wins)
• Loser backs off and tries different trunk
• Standard protocols ensure deterministic resolution

Just as circuits must be established before use, they must be formally released afterward. Circuit teardown is the final phase of the circuit-switching lifecycle, freeing resources for reuse.

Teardown initiation:

Teardown can be initiated by:

• Calling party on-hook — Caller hangs up first
• Called party on-hook — Callee hangs up first
• Network timeout — Extended silence or inactivity detected
• Network operation — Maintenance, reconfiguration, or failure response

The teardown message sequence:

Release message (REL) contents:

The REL message carries important information:

• Cause Code — Why the circuit is being released

  • Normal clearing (on-hook)
  • User busy
  • No answer
  • Call rejected
  • Network congestion
  • Facility failure

• Circuit Identification Code (CIC) — Identifies which trunk to release

• Release initiator — Indicates whether forward or backward REL

Resource deallocation sequence:

At each switch in the path:

1. Switch fabric disconnection — Internal cross-connect torn down
2. Trunk release — Outgoing trunk marked available
3. State cleanup — Call processing resources freed
4. CDR generation — Call Detail Record written for billing/analytics
5. Statistics update — Traffic counters adjusted

Abnormal teardown:

When components fail during a call:

Link failure:

• Adjacent switches detect loss of signal on trunk
• Both sides independently release resources
• If signaling is still possible, coordination messages exchanged
• If not, each side times out and releases locally

Switch failure:

• Upstream switches detect signaling timeout
• REL generated automatically after timeout
• Resources recovered without explicit coordination with failed switch

Orphaned circuits:

If teardown signaling is lost, circuits can become 'orphaned'—reserved but unused. Switches implement periodic audits:

• Compare active circuit table with peer switches
• Release any circuits not confirmed as active
• Typically runs every 5-30 minutes

We have comprehensively examined how circuits are established in circuit-switched networks. Let's consolidate the essential insights:

What's next:

Now that we understand how dedicated paths are established, we'll examine Resource Reservation—the mechanisms that actually allocate bandwidth, switch capacity, and transmission resources along the circuit. You'll learn about different reservation granularities, overbooking strategies, and how modern networks apply reservation concepts.