A Novel Genetic Service Function Deployment Management Platform for Edge Computing
Algorithm 1
Genetic algorithm-based service deployment algorithm for edge computing (GSDAE).
Input:
(1)
Information of edge computing network topology
(2)
Information of all MDCs.
(3)
Information of all application SFs.
Number of user requests. (2) K represents the number of functional service categories that need to be deployed. (3) Geneik denotes the number of k-type functional services that need to be deployed on the ith MDC. (4) The computing capacity required for SF i. (5) The number of required SFs.
Output: the SF deployment of assignment matrices for all services
(1)
The population of GSDAE:
(2)
k = 1;
(3)
While k < K + 1 do:
(4)
Create SFMatrix to store information related to the deployed SFs;
(5)
Decompose SFMatrix and generate the SFcapacity submatrix;
(6)
Transpose the SFcapacity matrix to generate SFcapacityT, which stores the computing capacity required for the deployed SF;
(7)
Normalise the Populationc matrix:
(8)
Generate CapMatrix to record the computing resources owned by each MDC;
(9)
Calculate Consumptionc matrix to record the number of computing resources consumed by using the Populationc deployment strategy in each MDC;
(10)
Check whether the deployment of SFs complies with the computing resources of the MDC;
Mij = CapMatrix - Consumptionc
(11)
if Mij >= 0 then
(12)
Geniik is suitable to be deployed on MDCi;
(13)
End
(14)
else
(15)
Geniik is unsuitable to be deployed on MDCi;
(16)
End
(17)
for each Geniik of MDC in the graph do
(18)
Use a greedy algorithm to calculate the overload of each MDC. Those MDCs with the lowest latency in the neighbouring networks and are not yet overloaded are initially chosen;
(19)
(20)
(21)
(22)
Return MDCi, k with the minimal latency and cost that is suitable for the deployment of SFi,k.
