Telephone operators are negotiating agreements with local telecom companies to host the traffic of their customers who are travelling abroad. Dispatching actual traffic between the different contracts that have been negotiated is key to benefit from the best deals, while meeting commitments. The optimum dispatch ratios can be counterintuitive for a human dispatcher and are far too complex (group deals and dynamic traffic adjustments). The challenges were:
- To provide a model that allows to capture all forms of contracts: commitments, thresholds with discounts (either on the incremental traffic or on its integrality), bundles of traffic (data, voice, sms), destinations, or companies, …
- To integrate an optimization module as a micro-service in the roaming-tools offer to telecom companies;
- To tackle large scale problem instances, over a one-or-two-years traffic projection with several interacting companies and many complex agreements between them;
- To handle the dynamic of the system combining current traffic realization with future projections, and accounting for previous traffic assignments and uncertainties on future traffic volumes;
- To offer simulation features that help companies in renegotiating their contracts.
The data-driven decision-support tool finds out which ratio of the traffic volume should be assigned to each roaming contract. The decisions are robust to uncertainties and can be re-optimized in view of actual traffic realization.
The tool is a key asset for yield management allowing the telecom companies to reduce significantly their roaming costs (from 25% to 50% compared to previously-used solutions).
This digital twin is the ideal tool to simulate the impact of different contracts configurations and to secure an edge in the negotiation process.