Paper #: 5.2.21
Title:  Indexing Techniques for Wireless Data Broadcast under Data Clustering and Scheduling

1. Problems
This paper addresses the improvement of wireless data broadcast techniques. The idea is to conserve power by using a broadcast system of disseminating data, lessening the power draw of a constant direct link between the requester and the provider.

2. New Idea and Strengths
-A hybrid indexing model combining the strengths of an index tree and a signature method of indexing data files through the broadcast. The idea is to use an index tree for the upper half of the structure, then cut directly to signature blocks. This will preserve almost all the benefits of the index tree, with almost the same quality as typical signature indexing.
-Clustering and scheduling are also considered when deciding upon an appropriate indexing method.

3. Weaknesses and Extensions
-This was a poor choice of paper, because the authors make frequent reference to ideas discussed in other papers with very little background. 
-The initial mathematical proofs for broadcast time are not compared against one another, so the benefits are not really obvious until the testing phase. They could have compared the estimated times to highlight what variables are going to provide the significant differences.
-A few concrete examples would have helped immensely in understanding what was going on in the paper. Early on arguments went like this: "non-clustered data can be divided into segments... which can be considered as clustered", which was confusing. The differences in the types of systems were not very clear, as they appeared to treat them all identically.
-It seems like for this system to be successful, the broadcast schedule and frequency would have to be known and must be local. In such a situation, it is not clear why a wired network would not be appropriate.