A Novel Time Delay Estimation Algorithm for 5G Vehicle Positioning in Urban Canyon Environments

Round 1

Reviewer 1 Report

This paper presents a time delay estimation for 5G ranging adapting a novel MUSIC algorithm for NLOS identification and mitigation.

Overall the motivation, related work, methods and results are presented in a clear and scientificly sound way.

This is why only minor revisions are suggested:

• For improving readability "Algorithm 1" should be put on one page, instead of dividing it on two pages.
• Section 3.3 "Unsupervised multipath estimation": Parameter setting for DBSCAN should be explained in more detail. Two relevant parameters are introduced, however only one ("radius") is set. Additionally, more explanations should be carried out concerning equation (34) (Why were parameters set like this?)

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This paper studies how to estimate time delay by applying NLOS signal cancellation. This problem is interesting, but the term of time delay is misleading because the actual propagation distance is not taken into account. It is better to rewrite this paper to explain the problem/method clearly. Following are some detailed comments.

 

1. Line 73, “In some cases, the signal strength of direct path is weaker than non-direct path in NLOS propagation scenario.”

It is better to explain in what kinds of scenarios these happen.

 

2. Most directly related references ([23]-[28]) are too old. Please add more recent literature.

 

3. The term of time delay is misleading. It is possible to estimate TDOA, but without the distance between TX and RX, how to estimate the absolute value of time delay caused by propagation?

 

4. Line 168, 

“The identification principle is: if the correlation peak with the minimum time delay is not the maximum correlation peak, it means that the signal has been propagated by NLOS.”

Generally, if the LOS component is not very too weak, it will have a peak, and its delay can be estimated. Please clarify why the NLOS cancellation is needed.

 

5. Line 277,

“When there is a NLOS path with strong power and the time delay is similar to the direct path, the correlation calculation before NLOS cancellation cannot distinguish this NLOS path, so it is impossible to cancel this NLOS path, and when the time delay is estimated with super resolution, this path will be mistaken as the direct path, leading to the decrease of the direct

path identification rate.”

This explanation is questionable. Intuitively, with higher time resolution, different paths become more separable, and probably more identifiable.

 

6. In Fig.8, the ranging error is the error of all signals, or only the ones that are determined to have direct paths? What are the false positive and false negative probabilities in direct path identification?

 

7. As for the experiments, what does the time delay 0 of Path 1 mean? The propagation delay is not considered at all? How are the ranging errors computed?

 

8. Because the estimations of (relative) delay and amplitude of multipath signals have errors, it will be interesting to see how accurate these estimations (delay and amplitude) are and the impact of their residual in the remaining signal.

 

Minor comments:

1. R in (32) is computed from S_NR in (27) instead of (12)?
2. Lambda in (34) is different from lambda in (30)?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

the paper 

 

1. Did the author of this paper knows how to compute " ANgle of Arrival"
2. Eq 1: What is mod2
3. Line 120.............In order to facilitate the verification of the algorithm in this paper, we simplify the generation method........ does it make sense
4. A lot of English grammatical mistakes. Also, the language of the paper is not technical. for example, line 133  In vector form, (5) is given by:
5. Figure 3: what is the output
6. mathematical modeling is greatly done in the paper. However, I don't understand either the path loss model, and factors were taken as constant. Please provide source

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors have answered all comments perfectly, except the most essential one about the delay (the third comment). 

In the proposed method, estimating the absolute delay involving “signal propagation time” requires the “transmitted signal” be available at the receiver side, and this requires that TX and RX be rigorously synchronized in time. But this seems impractical for vehicles, which in urban canyons cannot rely on GNSS for positioning, nor for time synchronization.

It is better to rewrite the introduction, and give a clear system model, e.g., what is assumed to be known, and how to apply the delay for positioning.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

All comments have been addressed and there is no other comment any more.