International Journal of Ad Hoc & Sensor Wireless Networks ========================================================== Special Issue for MSN05 PAPER REVIEW FORM ----------------- Title: Energy-Efficient Aggregate Query Evaluation in Sensor Networks Author(s): Zhuoyuan Tu, Weifa Liang Reviewer: Jie Lian Date To Be Returned: 10 Jan 06 ----------------------------------------------------------------------------- I. NUMERICAL RATING OF THE PAPER: --------------------------------------------------------------------------- Rating Category Your Rating Scale Definition --------------------------------------------------------------------------- Level of 5=Excellent Reader Interest |-----| 4=Very good | 2 | 3=Good |-----| 2=Fair 1=Poor --------------------------------------------------------------------------- Evaluation of Work 5=Excellent work and a major contribution and Contribution |-----| 4=Good solid work of some importance | 2 | 3=Solid work but marginal contribution |-----| 2=Marginal work and very minor contribution 1=Very questionable work and contribution --------------------------------------------------------------------------- Originality 5=Highly original work Novelty |-----| 4=Contains an original contribution | 3 | 3=Somewhat original |-----| 2=Variation of a known concept 1=Complete lack of original ideas --------------------------------------------------------------------------- Practical Utility 5=Excellent |-----| 4=Very good | 1 | 3=Good |-----| 2=Fair 1=Poor --------------------------------------------------------------------------- Quality of 5=Very good Presentation |-----| 4=Basically well written | 3 | 3=Readable |-----| 2=Needs considerable work 1=Unacceptably bad --------------------------------------------------------------------------- Technical 5=Very good Correctness |-----| 4=Good | 2 | 3=Flaws but easy to correct |-----| 2=Some flaws 1=Unacceptably bad --------------------------------------------------------------------------- Overall Rating 5=Strong accept (very good quality) (could be .5) |-----| 4=Accept (good quality) | 2 | 3=Accept if room (marginal quality) |-----| 2=Likely reject (low quality) 1=Definitely reject (has no merit) --------------------------------------------------------------------------- II. SPECIFIC COMMENTS & SUGGESTIONS (to be made available to the author(s)): ------------------------------------ Strength of the paper: The authors proposed two algorithms, LmNC and LmGaNC, to improve the lifetime of a sensor network. If we do not consider the construction cost, the routing trees constructed by using the two preceding algorithms can balance the energy consumption of sensors, and, therefore prolong the lifetime of a sensor network. Weakness of the paper: The two proposed algorithms in this paper have two significant drawbacks, which result in that the proposed algorithms are not practical, as follows: (1) The algorithms are centralized. (2) Both the algorithms require global knowledge of residual energy for all nodes in the network. The more detailed explanation is as follows. First, LmNC and LmGaNC are not distributed algorithms and can not been executed by each sensor node. Comparing with GaNC, which is a fully distributed algorithm, each node in GaNC only requires its one-hop neighbor information to construct the routing tree. However, it is not clear in LmNC and LmGaNC which node is selected to execute the algorithms. If the execution node is the base station, the question is that even though the base station can construct the routing tree, how each node in the network determines its parent. One possible way is that the base station floods the entire routing tree to the network such that all sensors can find their parent. However, this flooding operation has extremely high cost. Secondly, to execute LmNC and LmGaNC, the execution node needs to know the current energy levels for all nodes in the network. The energy consumption hidden behind this requirement is much higher than the energy consumption of a flooding operation, since the energy level of each sensor needs to be forwarded to the execution node via a multi-hop path and aggregation for the energy level information is not allowed. Even though the simulation results by using the constructed routing trees improve the lifetime of a sensor network, this improvement does not take the construction cost and other costs into account (costs associated to Drawbacks 1 and 2). If we consider the overall cost, it is reasonable to say that the proposed algorithms consume significantly more energy than the compared algorithms (GaNC, MST, and SPT). --------------------------------------------------------------------------- According to the preceding drawbacks, this paper makes very minor Contribution in the field of energy-efficiency query processing. Hence, the Overall rating is between likely reject and definitely reject. --------------------------------------------------------------------------- Other comments: To make the paper more valuable, the authors should revise the proposed algorithms such that they are fully distributed and the execution by each sensor only requires local knowledge. In other words, the authors should reduce the construction cost and other costs to a level which is comparable to the cost in GaNC (TiNA). Otherwise, the efficiency of constructed routing tree is meaningless. In addition, the authors should give more complete configuration of the simulation environment. Based on the given simulation configuration in the paper, the simulation can not be repeated by any third party. In the paper, the authors use the distance between two nodes as a factor to construct routing tree. However, the distance factor implicitly requires that all sensors in the network are equipped with power adjustable transmitters. This requirement brings additional costs: more expensive sensor nodes and synchronization among sensors since sensors can transmit message over a variable distance. III. CONFIDENTIAL COMMENTS (IF ANY) to be withheld from the author(s): ------------------------------------ Comments given under THIS item are for the EDITORS ONLY (to be withheld from the authors).