| 摘要 |
A method of operating a communications network, in which a primary learning process periodically (at time T, Fig 2) determines which routes communications sessions are assigned to in order that the session can be carried with a requested QoS, and a secondary learning process is used, in between instances of the primary learning process to check that the QoS constraints are still satisfied. If not, action is taken, for example using a load balancer to transmit the session over one or more further network routes. If the second process predicts that the QoS measure will be breached, the flow may be divided into two or more parts, each part being reallocated to a respective route (time T+y). The routes may have the same class of service that the session was originally assigned, or a different class of service that can still satisfy the requested QoS. If there are not suitable routes that can meet the QoS requirements, the session may have no change made, be terminated (time T+z), or remapped to routes which provide an improved quality of service. |
