Skip to main content

Routing Intent Algorithm

When you call someone in New Zealand from the UK you don't care how the calls happens.

The goal for a global telco is to leverage it's international backbone network (Ethereum) to route inbound traffic to it's intended destination at optimum cost by handing off (interoperability) to alternative carriers (Alts, L2s & L3s) they trust to deliver at a competitive cost and quality of service. A stable cost base is essential for implementing a pricing strategy that balances profits per route vs continued loyalty.

Automate routing case decisions to fulfill intentions to extract maximal value from a network of solution providers.

Reference Data​

Data Integrity is fundamental to making accurate data-driven decisions.

Routing Algorithm
  • Wholesale vs Retail Strategy
  • World Number Plan
  • Routing Case
  • Quality of Service SLAs
  • Network (Bandwidth) Capacity
  • Customer Records
  • Trade Deals
  • Time of Day
  • Regulations

Operational data​

  • Inter-carrier Destination Costs
  • Call Records
  • Aggregated Network Traffic
  • International Cost base
  • Inter-carrier Destination Pricing

Arbitrage​

During the early 2000s many Telco operators had not updated their systems to monitor or create routing decisions for international calls to mobile phones because this required collaboration and co-ordination across many departments and systems that had weak interfaces for transforming and exchanging critical data.

Adding to the complexity, a definitive standard for a world number plan does not exist. To operate effectively each of the following systems must share common reference data taken from a single source of truth.

  • Switches
  • Mediation
  • Pricing
  • Billing

Arbitrage opportunities arise against a Telco when it's reference data is outdated or is not aligned across business critical systems. As an example of this French mobile numbers begin +336 or +337 but in the early 2000s many international carriers had not updated their routing tables and reference data to reflect this. Therefore international operators based in France could route national mobile terminating calls over their international backbone to another country and pass on to another international carrier that was still charging fixed line rates for calls to France mobiles.

Network Management​

Use traffic profiling a growth projections plan where to invest in infrastructure.

  • Routes per switch (capacity)
  • Routing cases per switch (destinations)

Network Traffic Profiling​

Need to profile expected traffic based on:

  • Pricing relative to the market
  • Day of Week
  • Time of Day
  • Special events
    • Christmas
    • Ramadan
    • Olympics

Capacity Planning​

Physical cables from one switch to another. Where the other switch could be

  • International backbone
  • Domestic backbone
  • Another carrier

Quality of Service​

Product type examples are Retail and Wholesale.

Each product type must have a way to designate what level of service the customer had paid for either by route or A number matching.

Each Routing Case (Destination) had acceptable target KPIs. If a supplier falls below thresholds for a particular Destination and Product, add them to a blocklist.

  • Answer Seizure Ratio
  • Post Dial Delay

Trade Deals​

Prioritised volumes that had to be reached to avoid contract penalties.

  • Bilateral Agreements
  • Multi-lateral Agreements

Open-ended agreements with prices for Number Plans sent via spreadsheets.

See trade in telecommunications

Routing Algorithm​

In the dream scenario the network would be full but not overflowing capacity, with every destination perfectly priced to customers for maximum trading profits.

All traffic for each product type and destination, terminates with the first-choice selection with no routing overflow and all traffic terminated above benchmark quality

Pricing at the highest possible price to attract maximum customer volume without cause overflow as per point one.

Rules​

Sequence of logic to deliver routing plans that optimise profits against available capacity and acceptable user experience for call quality.

  • Overflow from first-choice selection is minimised
  • Use suppliers most valuable routes, then exclude
  • Backbone by customer type
  • No circular looping
  • Minimal overflow
  • Blocked destination suppliers
  • Trade Agreements
  • Dial Code Destination Matching
  • Dial Code Cost Blending

Preparation​

Gather and sanitize and standardize data.

Predicted Traffic Profile Projected Minutes by Routing Case code per Hour

  • Growth Trend
  • Historical Trend
  • Global Events
  • Pricing
  • Time of Day

Blend Rates

Load Blended Rate Profile for each routing case by mapping offering price, code overlap for each Time of Day

Routing Logic​

In an iterative process that exits after achieving a set variation threshold or maximum iteration limit.

  1. Lock in bilateral agreement deals with any special rules
  2. Find the order for each destination
  • run in order from highest volume destinations to least
  • do not consider blocked providers
  • get the first and second routing providers
  1. For each node on the network, against anticipated overflow setting, calculate the blended cost for each node in the backbone network add egress cost.
  2. Recalculate the best routing options for destination including backbone nodes
  3. Ensure no option for circular routing

Run the routing algorithm in iterations, after each pass of the routing order logic;

  1. Subtract the second choice cost from the first choice cost
  2. Order dataset by greatest saving
  3. Loop through to sum anticipated volume per provider for each destination/product profile based on volume demand (as)
  4. If a provider's safe outbound capacity is reached;
  • Remove provider with full capacity as an option
  • Add the provider and route to a list to be ignored as a routing option in the next routing iteration
  • Move next provider option forward

Cost Base and Pricing​

Goal: provide prices that balance demand with capability to deliver at predicted margins

Predict cost base of routing with a factor of safety to account for traffic overflow.

Establish the extremes of what the market with pay by analysing their price sheets.

Review objectives of relation with the provider in the big picture, their relative demand for minutes to destination.

Schema​