Optimal Margin Routing

Optimal Margin Routing is a method or system that allows backward feedback from the network to optimize routing decisions. The majority of systems available route calls from a matrix generated from least cost, commonly known as least cost routing. This is based on the general business rule that infers the best routing choice for the call is the cheapest one.

This system has many shortcomings, the most common being sending calls down a dead or congested route in order to try to achieve the best margin. Often these systems cause extra congestion, especially during busy hour, by trying to jam too many calls for the destination capacity.

The second problem with this method is that it fails to take into account that some customers may be willing to pay more in order to complete the call with better quality, or in many cases, to complete the call at all.

Optimal margin routing seeks to avoid these pitfalls by allowing the operator to define a set of rules based on quality as well as price. During periods of congestion, the system will penalize destination routes for clients who are requesting higher quality.

The downstream quality can be measured in many ways, but primarily by using moving averages to define expected call durations and completion rates.

Upstream feedback to customers can be provided in several ways. One method is to provide the customer with multiple calling tiers. During congested periods, the customer can be provided with standard no circuit messages via ISUP/H323/SIP messaging. The customer would then send another attempt in their next routing choice defined at a higher price. Other messages could be utilized to provide better information to the customer about congestion periods.

The operator can then attempt the call out a higher priced destination if necessary.

The system in general has a set of rules that define minimum and maximum margin based on customer or group of customer. This might vary based on time of day or other increments. The second portion of the decision process is a quality threshold.

The incoming customer has a quality threshold and the outgoing destination vendor has a beginning quality threshold. Both thresholds may vary dynamically, but the downstream threshold is subject to highest change. It can never go to 100% or 0% for very long as attempts always have to be made in order to maintain statistics.

These statistics would be kept for destination codes such as an area code, but in some cases would be kept and analyzed down to the individual exchanges or numbers.

A classic real world example would be the dynamic routing of fax or modem type calls. There are vendors who are unable to complete these calls. These calls are often individual numbers. A series of failures to a single number would generate a penalty that would force routing escalation to a higher quality tier.

A second example would be caused by a network outage or error due to bad programming or portability incompatibly where a paticular destination vendor is unable to complete the call.

A third example would be based on abandoned calls due to one way audio ( owa ) or ringback issues due to lost or misrouted messaging for call proceeding type messages. This is a common problem with digital phone systems from SS7, ISDN, and SIP ( 180/183 ) because there is not a required response end to end from the 'ringing' message.

Tweet

Share on Tumblr

Pin It