Adaptive call routing system
A system, method and apparatus are presented for the adaptive optimization of call quality and marginal profit in a telephony over data network, where the network contains a plurality of destinations, where calls are routable to each destination via a plurality of gateways in route to such destination, and where said gateways may serve more than one destination. The method comprises continually monitoring call traffic to each destination to first analyze whether the destination is currently volatile. If it is not, no action is taken. If it is, the method then reallocates the destination's traffic among some or all of the gateways in route to it.
1. In a telephony over data network system having a plurality of destinations, a method of predicting call traffic to a given destination for a given next time interval, comprising:
(a) evaluating historical call attempts to the destination at a first time interval immediately prior to the next time interval;
(b) evaluating the historical call attempts to the destination at a second time interval immediately preceding the first time interval;
(c) forecasting the call attempts for said next time interval utilizing a plurality of algorithms whose inputs are at least (a) and (b), said algorithms being both linear and nonlinear forecasting methods;
(d) selecting one of the forecasting results from (c).
2. The method of claim 1 , wherein said plurality of algorithms include at least one of linforecast, avgforecast, medforecast and dowforecast.
3. The method of claim 1 , wherein said step of selecting is done automatically by the system.
4. The method of claim 1 , wherein said step of selecting is done by a system operator.
5. The method of claim 3 , wherein a system operator can specify a factor by which the selected forecasting result is multiplied to correct for anomalous situations or events not adequately modeled by the forecasting algorithms.
6. The method of claim 4 , wherein a system operator can specify a factor by which the selected forecasting result is multiplied to correct for anomalous situations or events not adequately modeled by the forecasting algorithms.
7. In a telephony over data network system having a plurality of destinations, a method of predicting call traffic to a given destination for a given next time interval, comprising:
(a) providing a plurality of gateways each adapted for receiving calls and terminating at least one call among the calls to the given destination remote to the plurality of gateways;
(b) evaluating historical call attempts to the destination at a first time interval immediately prior to the next time interval;
(c) evaluating the historical call attempts to the destination at a second time interval immediately preceding the first time interval;
(d) forecasting the call attempts for said next time interval utilizing a plurality of algorithms whose inputs are at least (b) and (c), said algorithms being both linear and nonlinear forecasting methods;
(e) selecting one of the forecasting results from (d);
(f) changing call traffic volume passing through at least one gateway among the plurality of gateways based upon the selected results.
8. The method of claim 7 , wherein said plurality of algorithms include at least one of linforecast, avgforecast, medforecast and dowforecast.
9. The method of claim 7 , wherein said step of selecting is done automatically by the system.
10. The method of claim 7 , wherein said step of selecting is done by a system operator.
11. The method of claim 9 , wherein a system operator can specify a factor by which the selected forecasting result is multiplied to correct for anomalous situations or events not adequately modeled by the forecasting algorithms.
12. The method of claim 10 , wherein a system operator can specify a factor by which the selected forecasting result is multiplied to correct for anomalous situations or events not adequately modeled by the forecasting algorithms.