Review for the paper from IEEE Access Journal: Access-2018-16184

[Rizki Ginanjar]

Dear colleagues,

I would like to share my review regarding a paper which is considered to be published in IEEE Access Journal titled: Learning To Activate Relay Nodes: Deep Reinforcement Learning Approach   which can be seen on:

https://mc.manuscriptcentral.com/ieee-access?URL_MASK=87e7a22b92f142f1b5fcfeb4461637ae

The detailed review is summarized as follows:

This paper proposes a novel approach of the distributed algorithm in a multi-hop ad-hoc wireless sensor network to form the network autonomously using the reinforcement learning concept which enables the deployed nodes to learn about the manipulating its maximum transmission range of its current state to maximize the reward (which is the throughput and power consumption).

 

A.    Pros:

l   The contribution of the authors is thoroughly explained in the first chapter.

l   The problem explanation behind the proposed idea is clearly explained.

l   The illustration of the system overview (Fig.1) is really good to represent the whole component of the proposed idea.

l   The authors comprehensively surveyed the works related to the field of autonomous network formation in wireless sensor networks.

l   Good collection of references. Consisted of the most up-to-date and highly related paper to the proposed idea.

l   The exact process or algorithm of the proposed idea is explained in detail even for the assumption for the network parameters.

Cons:

l  The survey of the related works of the proposed idea is too long. It takes around 10-15 % part of the whole paper.

l  There is only one illustration about the overview of the implementation of the system which cannot cover the explanation of the system model explained on the third and fourth section.

 

B.    Summary:

This paper remarkably introduces a novel distribution algorithm for wireless sensor networks with the integration of deep learning system, based on the concept of reinforcement learning to manipulate the transmission range of each node. The explanation about the proposed idea is really good, no noticeable error on technical part, and highly recommended to be published in the future edition of the journal, based on the clarity and novelty of the paper.

 

Decision: ACCEPTED (WITH MINOR EDITS)