Advances, Systems and Applications
Table 2 Task partitioning and offloading decision optimization approaches based on the proposed taxonomy metrics
From: Task partitioning and offloading in IoT cloud-edge collaborative computing framework: a survey
Algorithms | End-user | Inter-dependent Subtask Model | Cost Model | Profiler | Decision Engine | Decision Granularity | Decision Strategy | Offloading Infrastructure |
|---|---|---|---|---|---|---|---|---|
Giurgiu et al. [46] | single user | module graph | latency, energy consumption | hardware, software, network | GO | module | online | cloud |
BBAP &MCGAP (Niu et al. [32]) | single user | object graph | latency, energy | hardware, network | GO | object | offline | cloud |
CATSA (Lakhan et al. [47]) | single user | module graph | latency | software, network | GO | module | online | cloud-edge |
MAUI (Cuervo et al. [20]) | single user | control flow graph | energy consumption | hardware, software, network | ILP | method | online | cloud-edge |
CloneCloud (Chun et al. [22]) | single user | control flow graph | latency, energy consumption | software, network | ILP | thread | offline | cloud |
MACS (Kovachev et al. [49]) | single user | service graph | energy consumption, QoS | hardware, network | ILP | module | online | cloud |
Yang et al. [50] | single user, multi-user | service graph | QoS, economic cost | hardware, network | MILP | module | offline | cloud |
ParGen (Wen et al. [31]) | multi-user | data flow graph | latency | hardware, software, network | ILP | module | offline | cloud-edge |
DTOS-LBBD (Alameddine et al. [51]) | single user | task graph | latency | hardware, software | MILP | task | offline | cloud-edge |
DCOS (Ding et al. [53]) | multi-user | program call graph | latency | hardware, software | ILP | task | offline | cloud-edge |
Li et al. [57] | single user | task graph | latency | software, network | MINLP | task | offline | cloud-edge |
QL-JTAR (Dab et al. [61]) | multi-user | task graph | energy consumption, latency | hardware, software, network | RL | task | online | cloud-edge |
Meng et al. [62] | single-user | data flow | energy consumption, latency | hardware, software | DRL | module | online | edge |
Su et al. [63] | multi-user | module graph | latency, QoE | hardware, software | RL | objext | offline | edge |
MA-DDPG (Lu et al. [64]) | multi-user | service graph | energy consumption, latency | hardware, software, network | DRL | module | online | edge |
Tang et al. [65] | multi-user | data flow | latency | hardware, software, network | DRL | task | online | edge |
FDRL (Cheng et al. [66]) | multi-user | data flow | energy consumption, latency | hardware, software, network | DRL | task | online | cloud-edge |
Wang et al. [67] | single-user | task graph | energy consumption, latency, QoS | hardware, software, network | DRL | task | online | cloud-edge |
TOLBO (Yan et al. [68]) | multi-user | data flow | energy consumption, latency | hardware, software, network | DRL | module | online | cloud-edge |