============= ======= Review 1 ======= > *** Contributions: What are the major issues addressed in the paper? Do you consider them important? Comment on the novelty, creativity, impact, and technical depth in the paper. In this paper, authors investigate the throughput-utility optimal CSMA protocol in general network topologies with low computation/communication complexities and worst-case delay guarantees for transmission jobs with diverse sizes. A two-timescale algorithm is proposed to dynamically make decisions in each time slot on 1) rate control, 2) link scheduling, 3) job transmission, and 4) job dropping. A CSMA-style random access control mechanism is integrated with the Lyapunov optimization framework for the algorithm design > *** Strengths: What are the major reasons to accept the paper? [Be brief.] Very strong theoretical analysis on the performance of algorithms proposed. > *** Weaknesses: What are the major reasons NOT to accept the paper? [Be brief.] As the paper is very dense, and most of the proofs are not included (put in a online report), it is very difficult to judge the correctness. It seems that the protocol is not pure CSMA protocol. It is more of a hybrid protocol. > *** Detailed Comments: Please provide detailed comments that will help the TPC assess the paper and help provide feedback to the authors. This is one of the better papers in my pile. The paper is very dense and seems to have a strong theoretical components. It seems that the protocol is not pure CSMA protocol. It is more of a hybrid protocol. It will be interesting to study a pure CSMA based protocol. It seems that the overhead of conducting the link scheduling is not included in the analysis of the performance. What are the internal relationships between control scheduling, and link scheduling? It will be better if some motivation and explanation are given for this. > *** Recommendation: Your overall rating (Please try giving as few borderlines as possible). B+ = (top 20% of reviewer's perception of all INFOCOM submissions, but not top 10%) weak accept (4) ======= Review 2 ======= > *** Contributions: What are the major issues addressed in the paper? Do you consider them important? Comment on the novelty, creativity, impact, and technical depth in the paper. This paper proposed a CSMA-based wireless communication protocol which can achieve arbitrarily close-to-optima throughput with worst delay guarantee. The authors argued that they are the first to provide worst-case delay guarantee in general network settings. The topic of this paper is important. And the authors have provided solid theoretical analysis with a clear presentation. However, this paper is an improvement based on two existing works. It is yet another work on the stack of papers studying CSMA protocols. Thus, this paper may not have a long-term impact in this area. > *** Strengths: What are the major reasons to accept the paper? [Be brief.] This paper is basically well-written. As a pure theoretical work, it has provided rigorous mathematical proof for some important issues. As it is claimed, the proposed protocol provides worst delay guarantee, and has low computation complexity and low communication overhead. > *** Weaknesses: What are the major reasons NOT to accept the paper? [Be brief.] This paper is an extension and combination of reference 14 and reference 15. The problem modeling, the optimization technique, and the scheduling algorithm are inspired by those two references. Therefore, the novelty of this paper is limited. I also consider the performance section to be weak. > *** Detailed Comments: Please provide detailed comments that will help the TPC assess the paper and help provide feedback to the authors. 1. The authors should demonstrate the connections and the key differences between this paper and those close-related references. 2. In the explanation of Algorithm 2, the concept of ˇ°mini-slotˇ± is not well-defined. 3. ˇ°INTENT messageˇ± should be introduced in detail when explaining Algorithm 2. 4. In Section V (Performance Analysis), the authors have provided detailed derivation, but the results and the insights are limited. Although they have proved the bounds, it is still not clear that how good the performance of this protocol can be. 5. In the paper, it has been proved that there is a tradeoff between the throughput and the worst delay which the proposed protocol can provide. However, this tradeoff has not been closely examined. > *** Recommendation: Your overall rating (Please try giving as few borderlines as possible). B = (top 30% of reviewer's perception of all INFOCOM submissions, but not top 20%) - accept if there is room (3) ======= Review 3 ======= > *** Contributions: What are the major issues addressed in the paper? Do you consider them important? Comment on the novelty, creativity, impact, and technical depth in the paper. This paper proposes a framework, which include a set algorithms, for resource allocation in CSMA-based wireless communication. The set of algorithms include rate control, link scheduling, traffic class selection, and admission control. The approach allows use of non-uniform size packets, multiple classes of services and is supposed to work for general network topologies. > *** Strengths: What are the major reasons to accept the paper? [Be brief.] In terms of problem formulation, this paper attempts to address a fairly complex problem, with many layers of algorithms. It attempts to provide algorithms that provides some notion of worst case delay and throughput utility that is within some gap from the optimal. > *** Weaknesses: What are the major reasons NOT to accept the paper? [Be brief.] Overall, many details in the paper is not well motivated and/or explain. It is also not clear how novel the work is beyond what has been done in [13] and [15]. While they may have add a more complex framework, the model does make a number of assumptions that raises the issue of practicality and usefulness. > *** Detailed Comments: Please provide detailed comments that will help the TPC assess the paper and help provide feedback to the authors. - It is mentioned that "A more practical model should allow the existence of transmission jobs consisted of one/multiple consecutive data packets, which should either be fully delivered to the desination or completely dropped." However, for a large data packet that have been fragmented, successful transmissions do not have to be consecutive. Also, there can also be end-to-end retransmission and some applications can tolerate losses. So the proposed model is actually too restrictive. - The model assumes that interference in binary and is known, which simplifies the problem signficiantly and may not be practical. - It is not clear why Equation (16) can be decoupled into four independent optimization problem. This is a key part of the approach and more explanation will be very useful. - The main result is stated in Theorem 4 which forunately is very complex and it is rather difficult to get any intuition on its behavior. For example, when the number of links increases, how scalable is the proposed approach based on theorem 4? Some other questions: (1) V is an user-defined parameter. What are the reasonable values of V and what is the trade-off involved in using a large V as the gap decreases with increasing V. (2) A small T provides better performance. However, say if T = s^{max}, then communicaiton overhead is incurred very frequently and brings down the performance significantly. These issues are not discussed and should be. - There is no experimental (simulation) result given in the paper. While this paper focuses the model, since the solution is optimal within some gap, it is still useful to validate their solution using some acutal paramaters or at least highlight scenarios wherby their solution will be effective. > *** Recommendation: Your overall rating (Please try giving as few borderlines as possible). C = (top 50% of reviewer's perception of all INFOCOM submissions, but not top 30%) - weak reject (2)