Table 2 Comparing the advantages and disadvantages of the proposed QL-based DQMAC and related protocols
From: A Q-learning-based distributed queuing Mac protocol for Internet-of-Things networks
| Â | Advantage | Disadvantage |
|---|---|---|
Our QL-based MAC | Optimal nodes in each contention achieved | Slow convergence in large environments |
Transmission delay does not increase. | Unsuitable for large networks | |
S-MAC [7] | Preserving energy | Time synchronization is required. |
X-MAC [8] | Low power listening | The receiver node needs to remain awake. |
B-MAC [9] | Can be scaled to a large network | Unable to provide multi-packet mechanisms |
ML-based MAC [19] | Does not need prior characteristics of the node | Unsuitable for large networks |
Sarsa [20] | Model-free algorithm | Slow convergence in large environments |
DLMA [21] | Faster convergence and more robust | High computational cost |
QL-MAC [22] | Adaptively schedule according to the traffic load | Increase energy consumption |
RL-MAC [23] | Individual traffic load for each node | The trade-off for latency and energy efficiency |