Reviewer A
Paper Summary (Please summarize the paper in your own words.)
The authors design and implement a system to take advantage of spatial reuse in WiFi. They want to do this without the need to share data across APs and without requiring tight clock synchronization. The main idea of their design is to allow STAs transmit concurrently by disabling CSMA. Then, the APs decode the interfering signals by learning the channel and the interference from others using interference reference symbols embedded into the desired signal. To achieve this they use an MMSE detector and a protocol that controls the duration of the first frame from each STA such that reference symbols from different STAs are received at different times from the APs.
Strengths (What are the main reasons to accept the paper? You may comment on the importance of the problems addressed, the novelty of the proposed solutions, the technical depth, and potential impact. Your overall rating should be supported by your review.)
The authors have implemented their idea in USRPs and have shown that it works in their limited experimental testbed setup.
Weaknesses (What are the main reasons NOT to accept the paper? Again, think about the importance of the problems addressed, the novelty of the proposed solutions, the technical depth, and potential impact. Your overall rating should be supported by your review.)
The authors motivate this work by pointing out that it would be too costly to synchronize APs such that they can do network MIMO (behave like a mega AP) and too costly to share data. While this is correct, the authors solution is also too costly and unlikely to work well in the dense large scale scenarios where network MIMO and their solution would make sense in the first place as they would lead in sizable rate increases.
Reasons why the proposed approach is not practical: Disabling CSMA creates a myriad of other issues. 802.11ax has a method to do so but this is not used in any practical setting because of this. Also, arbitrating the length of frames such that one may ensure that concurrent transmissions from STAs will result in preambles that do not overlap in time will not scale with many STAs. In some sense, while the authors potentially reduce the complexity at the PHY layer as compared to a network MIMO approach, they increase the complexity at the MAC layer and use ideas which are not backward compatible with today's WiFi practice.
The authors rely on channel reciprocity to estimate the channel but this works for the wireless channel only and not for the  electornics. So this will not work in practice unless hardware calibration is performed. This will not happen in the context of WiFi due to cost reasons.
The system needs the number of AP’s antennas to be greater than or equal to the number of active STAs. This is completely unrealistic for dense scenarios (which is the case of interest for solutions allowing concurrent transmissions as it is in those cases that the extra bandwidth is needed). This is completely unrealistic even if we assume that STAs have a single antenna only, not to mention the practical case where STAs have multiple antennas (even cellphones these days have multiple antennas).
Additional Comments (Additional comments (if any) that you would like to provide to the authors. Please do not repeat what you stated above. If none, leave the following blank.)
The related work is very limited. There is sizable prior work which attempts to create spatial reuse systems which do not need to implement network MIMO, by avoiding tight synchronization and data sharing. The authors may want to study this work and more carefully explain the difference and contribution of their approach.
Also, the authors should think more about whether their proposal is practical in the context of today's WiFi networks, as this is the application they target.
Confidential Comments to TPC (if any) (You may provide any confidential comments about the paper. If the paper has violated the INFOCOM double blind policy, you should provide specific details here. This area is not visible to the authors.)
I have see 3 papers with the exact same ides, presumably from the same authors. I have delegated 2 of those papers to a graduating PhD student of mine (he has defended his thesis already). The authors apply the same idea in 3 applications. In my opinion the idea is not practical in neither of the three, though the level of inapplicability is probably  bit different. I am only providing the review for this one. FYI:

This paper:
TCCI: Taming Co-Channel Interference for Wireless LANs with Limited Cooperation
The other two papers:
DM-COM: Combining Device-to-Device and MU-MIMO Communications for Cellular Networks
UD-MIMO: Uplink Distributed MIMO for WLANs


Reviewer B
Paper Summary (Please summarize the paper in your own words.)
This paper presents TCCI, a new scheme to enable concurrent transmissions in WLANs by leveraging APs’ multiple antennas for co-channel interference management. This scheme consists of a new detection and beamforming method for the Aps. The new detection method, an approximate-MMSE detector, decodes the desired signals by leveraging the reference symbols embedded into desired signals in the absence of channel knowledge. The beamforming method used in the downlink is designed taking advantage of channel reciprocity and the approximate-MMSE detector computed in the uplink.
Strengths (What are the main reasons to accept the paper? You may comment on the importance of the problems addressed, the novelty of the proposed solutions, the technical depth, and potential impact. Your overall rating should be supported by your review.)
Paper is well written and organized and the authors have built a prototype of TCCI on a wireless testbed.
Weaknesses (What are the main reasons NOT to accept the paper? Again, think about the importance of the problems addressed, the novelty of the proposed solutions, the technical depth, and potential impact. Your overall rating should be supported by your review.)
1. The paper claims that TCCI, the scheme presented by this paper, needs very limited cooperation among the Aps and therefore is more amenable to practical implementation in real networks. However, the practicability of this scheme is questionable.
Firstly, this paper only considers three simple WLAN scenarios, without mention how to extend to more complex scenarios in real life.
Secondly, the scheme requires that the number of antennas of AP should not be less than the number of active STAs, which can be satisfied with the case of small WLAN network. However, This is obviously unrealistic for dense scenarios.
Thirdly, the scheme requires STAs only have one antenna, which does not conform to the reality that may devices have multiple antennas.
2. The paper also claims that TCCI does not require data sharing, channel knowledge, or fine-grained synchronization for the network with an approximate-MMSE detector and the beamforming taking advantage of channel reciprocity. However, all of these operations require a certain amount of computation in MAC layer. Is TCCI really more effective than MIMO?
3. The interference management techniques introduced in RELATED WORK section is too brief, and more detailed and relevant work needs to be added.