A Survey of IoT Stream Query Execution Latency Optimization within Edge and Cloud

Table 4

Query execution models within the context of their strengths and weaknesses.

Query execution models	Strengths	Weaknesses
Cloud-based query processing	(i) Abundant compute and storage resources; (ii) Optimized processing delay	(i) Longer networking delays; (ii) In-affordable for latency-sensitive queries
Geo-distributed query processing	(i) WAN-aware query computation; (ii) Optimized networking delay; (iii) Network-awareness incorporation in QEPs	(i) WAN-data transmission cost
Edge-cloud-based query processing	(i) Partial computations on edge; (ii) Reduced WAN usage cost; (iii) Reduced networking delay	(i) Limited and resource-constrained nature of edge devices; (ii) Network-compute resource heterogeneity
Edge-based query processing	(i) Complete query execution on edge; (ii) Hop count reduction between data sources and query execution node; (iii) Reduced networking delay	(i) Unsuitable for the complex type of queries; (ii) Resource heterogeneity