Table 2 Key parameters of the simulation environment for training and testing the delay shift agent.
From: Decentralized queue control with delay shifting in edge-IoT using reinforcement learning
Parameter | Value |
|---|---|
Number of IoT devices | 1250 (50 per each of the 25 edge nodes) |
Task arrival rate | 10–20 tasks per minute (Poisson distribution with ±10% fluctuations) |
Number of edge nodes | 25 |
Node CPU frequency | 2–3 GHz |
Node RAM | up to 8 GB |
Node power consumption | up to 30 W, 180 Wh (power budget) |
Fog latency | 250 ms |
Network topology | BA graph (attachment parameter m = 2) |
Channel bandwidth | 1–20 Mbps |
Channel latency | 10–150 ms |
Simulation step/episode duration | 1 s/200 steps |
Random number generator | fixed seed (seed = 42) |
Input datasets | Orange D4D, Intel Lab Sensor Data |
Programming language | Python 3.10 |
Libraries used | TensorFlow 2.11, NumPy 1.24 |
Hardware platform | AMD Ryzen 7 5800X, 64 GB RAM |
