Abstract: In the design of embedded system, it needs for semantic retrieval task scheduling, routing conflict, because the link disjoint routing protocol, resulting in semantic retrieval code in the link load of semantic integration, scheduling and performance is not good. This paper presents a semantic retrieval model covariance fusion based on link separation, firstly analyzes the routing link embedded in the conflict, the mapping relationship between ontology and semantic equivalence model of heterogeneous, followed by the key words of each document set is processed, each subject automatic attribute added to form the background concentration, improve the data mining and semantic retrieval ability, to realize the semantic link disjoint routing conflict covariance fusion retrieval model. The experimental results show that, using this method can effectively realize the separation of routing link embedded under the conflict, realizing the link of covariance fusion, semantic data transmission embedded system can improve the data throughput rate, it has good application value in the design and communication of embedded system.