Method and system for network slice allocation

1. A method of allocating network slices of a network infrastructure, the method comprising:

receiving a network slice request for network resources of the network infrastructure in a form of a network slice, the network slice request including a service level agreement (SLA) and an associated payoff;

determining whether to accept the network slice based on whether it is expected that a utility function will be better served by accepting the network slice request or waiting for a further network slice request, wherein the determination takes into account SLA multiplexing properties and a prediction of whether a further SLA of the further network slice will provide a higher or lower associated payoff;

determining whether the SLA would be fulfilled prior to allocating the network slice;

allocating the network slice; and

installing the network slice in the network infrastructure.

2. The method according to claim 1 , wherein a value iteration algorithm or an adaptive algorithm is used to determine whether the utility function will be better served by accepting the network slice request or waiting for a further network slice request.

3. The method according to claim 2 , wherein the adaptive algorithm is used to determine whether to accept the network slice request, the adaptive algorithm being based on a Q-learning framework.

4. The method according to claim 3 , wherein the adaptive algorithm maintains Q-values which are iteratively updated based on new events and uses temporal difference learning, and exploration and exploitation procedures.

5. The method according to claim 4 , wherein past information including previous payoffs and resource utilization based on other SLAs having requirements comparable to the SLA is considered to determine whether to accept the network slice request.

6. The method according to claim 2 , wherein the value iteration algorithm is used to determine whether to accept the network slice request, the method further comprising determining transition probabilities based on a Semi-Markov Decision Process and normalizing transition times as multiples of a faster, fixed transition time so as to form a discretized Markov chain.

7. The method according to claim 1 , wherein the utility function is overall network resource utilization.

8. The method according to claim 1 , wherein the utility function is overall system revenue.

9. The method according to claim 1 , wherein the method is performed by a network slice controller disposed in a mobile network architecture, the network slice controller utilizing a network exposure function (NEF) for receiving the network slice request from an infrastructure tenant.

10. The method according to claim 1 , wherein the network slice is elastic, and wherein the determination of whether to accept the network slice includes determining whether it is expected that waiting for another network slice request for a network slice that is inelastic will better serve the utility function.

11. The method according to claim 1 , wherein past information including previous payoffs and resource utilization based on other SLAs having requirements comparable to the SLA is considered to determine whether to accept the network slice request.

12. A network slice controller for allocating network slices of a network infrastructure, the network slice controller comprising one or more processors which, alone or in combination, are configured to provide for performance of the following steps:

receiving a network slice request for network resources of the network infrastructure in a form of a network slice, the network slice request including a service level agreement (SLA) and an associated payoff;

determining whether to accept the network slice based on whether it is expected that a utility function will be better served by accepting the network slice request or waiting for a further network slice request, wherein the determination takes into account SLA multiplexing properties and a prediction of whether a further SLA of the further network slice will provide a higher or lower associated payoff;

determining whether the SLA would be fulfilled prior to allocating the network slice;

allocating the network slice; and

installing the network slice in the network infrastructure.

13. The network slice controller according to claim 12 , being further configured to use a value iteration algorithm or an adaptive algorithm to determine whether the utility function will be better served by accepting the network slice request or waiting for a further network slice request.

14. The network slice controller according to claim 13 , wherein the adaptive algorithm is used to determine whether to accept the network slice request, the adaptive algorithm being based on a Q-learning framework, and wherein the adaptive algorithm maintains Q-values which are iteratively updated based on new events and uses temporal difference learning, and exploration and exploitation procedures.

15. The network slice controller according to claim 14 , wherein past information including previous payoffs and resource utilization based on other SLAs having requirements comparable to the SLA is considered to determine whether to accept the network slice request.

16. The network slice controller according to claim 12 , wherein the network slice controller is disposed in a mobile network architecture, the network slice controller utilizing a network exposure function (NEF) for receiving the network slice request from an infrastructure tenant.

17. The network slice controller according to claim 12 , wherein past information including previous payoffs and resource utilization based on other SLAs having requirements comparable to the SLA is considered to determine whether to accept the network slice request.