=========================================================================== ICNP15 Review #48A Updated 2 Jun 2015 11:13:17am CDT --------------------------------------------------------------------------- Paper #48: Wi-FM: Resolving Neighborhood Wireless Network Affairs by Listening to Music --------------------------------------------------------------------------- Overall merit: 4. Accept Reviewer expertise: 2. Some familiarity Novelty: 4. New contribution ===== Paper summary ===== Paper shows that the digital data signal accompanying FM radio broadcast can be used as an out-of-bound mechanism to support synchronized time-slotting for WiFi networks across administrative bounds. Paper further demonstrates that mutual interference on a given WiFi channel can be reduced substantially by combining this synchronization facility with their proposed TDM scheme. Further this scheme requires no explicit coordination across WiFi nodes, utilizing distributed, independent broadcast choices at each WiFi transmitter. ===== Strengths ===== Neat idea. PoC implementation. Reasonable results and generally clear presentation. ===== Weaknesses ===== Several cryptic/unexplained points/design choices. Unclear how impactful ideas truly will be. Would be more impressive if authors attempted to tackle how their mechanism should interact with the channel selection problem. ===== Comments for author ===== Overall I liked your paper, very neat idea and kudos on the implementation & evaluation. Several items need to be fixed - Section 3, equation 2 is buggy. The index of summation is improperly written with 'i' below the summation symbol and 'B_k' above. You should instead have 'i \in B_k' at the below the summation symbol. - It would appear that the critereon from equation 2 might result in hysterisis. Can you comment on why this won't be the case? (or might it?) - Please make the distinction that this work reduces contention in on a single WiFi channel and treats the channel selection problem as orthogonal (unless I've substantially misunderstood, in which case this really needs clarification). - in section III.B, second paragraph, please clarify what is meant by "WiFM keeps track of the amount of time in which there are packets in the outgoing buffer of a node. If this time is in excess of 2 seconds..." --- does this mean that 'if the buffer is continuously not empty for a 2-second period' or 'if any packet stays in the buffer for 2-seconds' or something else? Likewise, is this per flow (5-tuple) or across all flows that the sender is sending? If this condition is met once, is the classifcation of sender as heavy made immediately or is this measurement repeated a couple of times? Is the decision ever reassessed? If so, when? If not, why? - In IV you write that "TO maximize the responsiveness to changes in state at the queuing discipline level, we decrease the available buffer size on each card". I found this cryptic. Please elaborate. - In IV you note that you disable RTS/CTS on APs. Is this a reasonable thing to do? It seems to make things easier for your scheme, which could be construed as inflating the performance gain you achieve. I'd like to see a direct comparison of a your system with one that uses the standard setup, including RTS/CTS on the APs. - In V.A you only sample stations in the bottom half of the FM radio range. Why only the bottom half? =========================================================================== ICNP15 Review #48B Updated 3 Jun 2015 4:49:08am CDT --------------------------------------------------------------------------- Paper #48: Wi-FM: Resolving Neighborhood Wireless Network Affairs by Listening to Music --------------------------------------------------------------------------- Overall merit: 4. Accept Reviewer expertise: 2. Some familiarity Novelty: 4. New contribution ===== Paper summary ===== This work address using FM capability of a mobile device to improve performance of wireless connectivity. ===== Strengths ===== - The paper describes the details quite well. - It's well written. ===== Weaknesses ===== - It doesn't discuss what happens if an FM frequency is in service by a radio station. Secondly, now-a-day at least in large cities, almost all FM frequencies are used up. So, in practice, will there be much gain due to limited FM frequency available? - It's mostly an implementation exercise (nothing wrong with that) ===== Comments for author ===== This is an interesting work on the using of FM for wireless connectivity to improve performance. The work is described well. While I like the approach and the presentation, it doesn't discuss what happens if an FM frequency is in service by a radio station. Secondly, now-a-day at least in large cities, almost all FM frequencies are used up. So, in practice, will there be much gain due to limited FM frequency available? =========================================================================== ICNP15 Review #48C Updated 5 Jun 2015 12:33:36am CDT --------------------------------------------------------------------------- Paper #48: Wi-FM: Resolving Neighborhood Wireless Network Affairs by Listening to Music --------------------------------------------------------------------------- Overall merit: 4. Accept Reviewer expertise: 2. Some familiarity Novelty: 4. New contribution ===== Paper summary ===== This paper proposes using FM radio signals for harmonising WiFi transmissions amounting neighbouring devices. The authors note that FM radio chips are now bundled together with WiFi and Bluetooth chips, making the FM receiving pervasive among smartphones. The authors show that they can use the RDS signal from FM transmissions to harmonise neighbouring WiFi devices relative to this RDS signal. This allows more efficient sharing of the wireless channel especially when multiple devices have long-lived heavy flows. ===== Strengths ===== I really liked the idea, and its well presented, and backed up with experimentation. This is a very nicely written paper. Now, I don't really work with lower layer technologies, so may be my excitement might be partly due to my ignorance, but I did like this paper very much. I would have gone with a "5" rating but tempered it down to "4" because of I 'm not an expert in this area. ===== Weaknesses ===== None that I can think of. ===== Comments for author ===== I really enjoyed this paper. It's a really cool idea, and the paper makes a very strong case for FM signal based harmonisation for improving WiFi performance. The synchronisation is achieved with little costs and zero overhead. I also commend you for making the presentation so accessible. This was a treat to read! =========================================================================== ICNP15 Review #48D Updated 18 Jun 2015 5:24:08pm CDT --------------------------------------------------------------------------- Paper #48: Wi-FM: Resolving Neighborhood Wireless Network Affairs by Listening to Music --------------------------------------------------------------------------- Overall merit: 2. Weak reject Reviewer expertise: 4. Expert Novelty: 2. Done before (not necessarily published) ===== Paper summary ===== The paper uses certain FM signal features to synchronize/schedule WiFi transmissions. This manages interference better. The paper uses an SDR implementation to demonstrate modest performance improvements. ===== Strengths ===== Using FM signals to synchronize is novel. The authors are able to achieve this by very low cost SDR devices (DTV dongle and open source software). The quality of synchronization has been analyzed well. ===== Weaknesses ===== The actual scheduling part is quite ad hoc and presented poorly. Frankly I did not understand very well what is being done in Sec IIIB. See more below. ===== Comments for author ===== - The synchronization part is well done. But the scheduling part is poorly presented. - I could not understand very well how the scheme really works. There is no centralized scheduler. But then exactly how is the amount of concurrent interfering transmissions minimized? What is the scheduling objective really? - What happens when there is a mix of Wi-FM and regular nodes? - The testbed is really small. - There is a significant amount of TDMA literature already for WLAN. But there is no real comparison with anything else. =========================================================================== ICNP15 Review #48E Updated 30 Jun 2015 2:58:23am CDT --------------------------------------------------------------------------- Paper #48: Wi-FM: Resolving Neighborhood Wireless Network Affairs by Listening to Music --------------------------------------------------------------------------- Overall merit: 4. Accept Reviewer expertise: 2. Some familiarity Novelty: 4. New contribution ===== Paper summary ===== TPC summary review: The reviews for the paper reflect the general enthusiasm about the novelty of the approach and was the basis for accepting the paper. There were clear reservations, however, about the treatment of transmission scheduling including its ad-hoc nature and poor explanation. It is expected that the authors will take this feedback into account and work to significantly improve the scheduling portion of the paper in the final version.