REVIEWER A
----------

This paper covers the area of bandwidth estimation in IPv6 packet networks.
This paper is divided in two parts. First, it proposes a timestamp option
for IPv6 protocol using hop-by-hop extension header. Second, it shows that the
proposed timestamp option can be used to estimate capacity and available
bandwidth of end to end path.  

The timestamp proposal requires that each router in the path timestamp
the packet with sufficient precision and put the precision information in the
packet. The bandwidth estimation (both capacity and available bandwidth
estimation) can then be done by simple modifications to existing bandwidth
estimation methodologies. The modification is to use timestamps in the
IPv6 header instead of that collected at the end hosts. 

The paper in its current form lacks tutorial content. The paper is basically a
proposal for incorporating timestamp option in IPv6 and shows its usefulness in
bandwidth estimation. This contribution is not suitable for publication in
IEEE network and would be more suitable for IETF forums. 

Besides not being suitable for IEEE network, this paper could be improved in
the following directions. Authors do not provide sufficient proof that existing
bandwidth estimation techniques are not consistent and accurate, which is one
of their main motivations to add timestamp option. Scenario illustrated in
figure 1 only affects a subset of tools that relies on dispersion of packets
for either capacity or available bandwidth estimation.  This problem has been
extensively studied by Dovrolis et al. (Packet Dispersion Techniques and
Capacity Estimation) in context of capacity estimation. Furthermore, there are
methodologies for available bandwidth estimation which use on one way delay
trend and not rely on the dispersion of probing packets. Following papers by
Dovrolis et al. provide nice overview of such bandwidth estimation
methodologies - "Ten Fallacies and Pitfalls in End-to-End Available Bandwidth
Estimation" and "Bandwidth Estimation: Metrics, Measurement Techniques, and
Tools"

Second place where paper can be improved is in section 5 describing the
proposed bandwidth estimation methodologies. As different routers add
timestamps in a packet, the packet size would increase, making s(f) in eq (3) a
function of the hop count i. It is not clear why the packet size is fixed in eq
(3).  Additionally, the bandwidth estimation using timestamp option in IPv6
would still suffer from layer-2 issues that were discussed in detail in paper
by Prasad et al. (The Effect of Layer-2 store-and-forward devices on per-hop
capacity estimation). This should be clarified in the paper.