FastFlow: Efficient Scalable Model-Driven Framework for Processing Massive Mobile Stream Data

Algorithm 3

Global optimizer.

Input: <<operationModel>>: a set of operation list

: cost of cpu, memory and bandwidth for special operation

: weight of different resources

Output: metaTopology: the optimal execution topology

Step  1. Initialize and split basic operators

(1) Initialize (0-1)-matrix OperationMatrix(, )

(2) Extract basic operators com(operationModel) from <<operation Model>>

Step  2. Set similarity of multiple queries

(3) calculate pairwise JaccardSimilarity(Q1, Q2)

(4) <Query, Query>   order pairwise queries by their similarity score

Step  3. Local optimizer

(5) localoptimalTopology    <>

(6) localoptimalTopology    Meta-Model Local Optimize(<<operationModel>>, , )

Step  4. Global optimizer

(7) metaTopology    <>

(8) for all Q1, Q2 in <Query, Query>  do

(9) if Q1, Q2 not in metaTopology  then

(10) metaTopology   Merge Q1, Q2 and keep global semantic

(11) end for

(12) return  metaTopology