kumoh national institute of technology
Networked Systems Lab.

Ade Pitra, Dong-Seong Kim, and Jae-Min Lee, "Sensor Failure Recovery using Multi Look-back LSTM Algorithm in Industrial Internet of Things", IEEE ETFA 2020. (A)
By : Ade Pitra
Date : 2020-07-08
Views : 148

REVIEWER 1
C. Clarity of presentation:
English grammar and spelling are proper ------------------------------------------ [3 - I agree]
Mathematical symbols and equations are easy to understand ------------------------ [3 - I agree]
Figures and tables are well constructed and informative -------------------------- [3 - I agree]
The paper is well organized ------------------------------------------------------ [4 - I strongly agree]
Considering the issues above, the paper is readable ------------------------------ [3 - I agree]

T. Technical innovation and relevance
The authors cite other relevant publications ------------------------------------- [4 - I strongly agree]
Authors describe relevance of work to the research field ------------------------- [3 - I agree]
The authors apply sound technical approaches ------------------------------------- [3 - I agree]
New ideas are convincingly and logically described ------------------------------- [4 - I strongly agree]
Results are convincing ----------------------------------------------------------- [3 - I agree]
Considering the issues above, this work should be presented --------------------- [3 - I agree]

Comments:
The paper describes a technique to recover missing sensor
data and uses a significant data set. Preliminary results
are presented and briefly discussed.



REVIEWER 2
C. Clarity of presentation:
English grammar and spelling are proper ------------------------------------------ [1 - I disagree]
Mathematical symbols and equations are easy to understand ------------------------ [3 - I agree]
Figures and tables are well constructed and informative -------------------------- [2 - I am neutral]
The paper is well organized ------------------------------------------------------ [3 - I agree]
Considering the issues above, the paper is readable ------------------------------ [2 - I am neutral]

T. Technical innovation and relevance
The authors cite other relevant publications ------------------------------------- [3 - I agree]
Authors describe relevance of work to the research field ------------------------- [2 - I am neutral]
The authors apply sound technical approaches ------------------------------------- [2 - I am neutral]
New ideas are convincingly and logically described ------------------------------- [2 - I am neutral]
Results are convincing ----------------------------------------------------------- [2 - I am neutral]
Considering the issues above, this work should be presented --------------------- [2 - I am neutral]

Comments:
The paper would benefit from a further language review. In
parts it is hard to understand the point the authors are
trying to make due to incorrect choice of words and grammar.

The mention of relevant related works in the introduction is
quite confusing as the authors mention both research dealing
with the problem at hand, but also research that merely
deals with wireless communication in IIoT. My recommendation
would be to concentrate on related works that deal with the
same problem that the paper is trying to solve, and then
contrast the paper's contribution to existing solutions (or
the lack of those). Preferably in the section on related
works. The argument for the relevance of this work in the
introduction should be stronger linked to actual problems
(e.g. an application?) that would benefit from or even rely
on the solution.

Some more explanations for the choice of parameter settings
in the different algorithms would be beneficial, so one can
actually make a good analysis of why certain algorithms are
better than others. I am missing some hyperparameter
optimizations and also a more thorough analysis av the
results, not merely a description of them.



REVIEWER 3
C. Clarity of presentation:
English grammar and spelling are proper ------------------------------------------ [1 - I disagree]
Mathematical symbols and equations are easy to understand ------------------------ [3 - I agree]
Figures and tables are well constructed and informative -------------------------- [3 - I agree]
The paper is well organized ------------------------------------------------------ [3 - I agree]
Considering the issues above, the paper is readable ------------------------------ [2 - I am neutral]

T. Technical innovation and relevance
The authors cite other relevant publications ------------------------------------- [3 - I agree]
Authors describe relevance of work to the research field ------------------------- [3 - I agree]
The authors apply sound technical approaches ------------------------------------- [2 - I am neutral]
New ideas are convincingly and logically described ------------------------------- [2 - I am neutral]
Results are convincing ----------------------------------------------------------- [3 - I agree]
Considering the issues above, this work should be presented --------------------- [3 - I agree]

Comments:
The paper proposes to use a Recurrent Neural Network to detect sensor faults and attempt to recover missing data in an IOT application.
A few questions:
- How did you choose the number of nodes in the MLP and LSTM?
- how can the dependence on the season be modeled? Is the date and time part of the input?
- what is the impact of the SARIMAX elements which are not part of ARIMA, and hence are not modeled by your LSTMs? If you get a good loss function value for the part that matters, but ignore other errors, you may not be actually doing a good job overall. Please clarify.



REVIEWER 4
C. Clarity of presentation:
English grammar and spelling are proper ------------------------------------------ [3 - I agree]
Mathematical symbols and equations are easy to understand ------------------------ [2 - I am neutral]
Figures and tables are well constructed and informative -------------------------- [3 - I agree]
The paper is well organized ------------------------------------------------------ [3 - I agree]
Considering the issues above, the paper is readable ------------------------------ [3 - I agree]

T. Technical innovation and relevance
The authors cite other relevant publications ------------------------------------- [2 - I am neutral]
Authors describe relevance of work to the research field ------------------------- [2 - I am neutral]
The authors apply sound technical approaches ------------------------------------- [3 - I agree]
New ideas are convincingly and logically described ------------------------------- [3 - I agree]
Results are convincing ----------------------------------------------------------- [2 - I am neutral]
Considering the issues above, this work should be presented --------------------- [2 - I am neutral]

Comments:
An algorithm to recover failure sensor data in real-time
is proposed. In the current factory automation trends,
each device senses the information directly, where humans
are no longer involved in huge part of the task. The
ability of the system to detect the failure in the device
is one of the crucial things as well as the ability to
recover the data of the device when failure happens.

(i) LSTM can be put into Background.
(ii) The theory of your contribution seems weak. Please
give theoretical explanation.



The manuscript is marked as ACCEPTED.