A Transformation Technique for Scheduling Broadcast Programs of Multiple-Item Queries

In wireless environments, many mobile users may raise queries for accessing multiple data items, e.g., stock information or traffic condition, simultaneously. Most queries are identical. That is, many users may have an interest in some popular data items. Because data broadcast can offer unlimited users shareable information at the same time, a broadcast server is usually employed to disseminate all the data items periodically. Intuitively, popular data items should be scheduled and transmitted to users more efficiently than ordinary ones, so these users can thus save access time. To improve customer satisfaction, the author considers a broadcast program scheduling problem in such an environment and aims to minimize mobile users’ worst access time as well as their battery power consumption by generating near-optimal broadcast programs. The author provides theoretical analysis as a foundation of mapping the problem to another domain (i.e., from unit item to unit fragment) and this transformation makes the problem easy to solve. Moreover, an O(N logN) algorithm is proposed for this NP-hard problem. Finally, experimental results show that access time can be reduced by carefully scheduling broadcast programs. It suggests that other similar optimization problems can be solved similarly. DOI: 10.4018/jghpc.2012100104 International Journal of Grid and High Performance Computing, 4(4), 52-67, October-December 2012 53 Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. be denoted by a query set, e.g., q1={d1, d4, d12}, where d1, d4, d12 are the keys of the desired data items. These users can download the index information about q1 first and determine where the desired data items locate. Then their mobile devices (e.g., PDA) can enter the doze mode to wait for desired items and later return to the active mode to access them. Note that the battery power that mobile devices consume in the active mode is much more than that they do in the doze mode (Shin, 2011; Waluyo et al., 2011). Therefore, it is important to design a short broadcast program constrained by that all items in a query set should be organized together. The reasons are as follows. First, a short program means less access time. That is, it can improve customer satisfaction. Second, placing all data items in each qi together means that only one index entry for qi is enough. This is because data items in qi will be organized into a consecutive broadcast subprogram. And thus only a small index structure is needed. This signifies we can consume less power to locate a query set. In such an environment, customer satisfaction and power consuming are measured by two common metrics: tuning time and access time. Tuning time (Huang & Huang, 2010; Shin, Wu, & Chen, 2011; Waluyo et al., 2011; Yang, Shi, Wu, Gao, & Zhong, 2011) is the time spent by a mobile user only in the active mode before downloading all the desired data items. On the other hand, access time (Huang & Huang, 2010; Jea & Wang, 2010; Jea et al., 2008; Wang & Jea, 2009) is the waiting time spent by the user before receiving all desired data items in both modes. Hence, access time is an upper bound of tuning time. Moreover, a program scheduling that generates the optimal tuning time does not mean it also ensures the optimal access time, and vice versa. Few of past schemes paid attention on both issues for multiple-data-item access, so we intend minimizing both of them. In this paper, we aim to minimize the worst access time by generating a short broadcast program for all users. For example, consider there are two query sets q1={d1, d4, d12} and q2={d4, d7, d12}. The broadcast program π1=[d1, d4, d12, d7] as well as π2=[d1, d4, d12, d4, d7, d12] is of the same function, since both can periodically disseminate the same information. But the former is much more favorable than the latter. This is because the worst access time for π1 is four (i.e., four items) and it is also the optimal. So these overlapped items, e.g., d4, d12, should be well-organized. This is the key to the best scheduling. Although our goal is to minimize access time, we do not achieve it at the cost of tuning time. In the above example, if we organize the broadcast program like [d1, d4, d12, d7], we need only one index entry for this query set q1 (i.e., to inform users where d1 locates is enough) and one entry for q2 (i.e., where d4 is). And thus we will spend only a little tuning time when locating q1 or q2 by reading a small index structure. The reason is that an index structure for indexing all data items in each query set is very likely to become more complicated and larger e.g., a B+ tree or a Huffman tree. That is, it will cause longer tuning time (i.e., consume more power) for locating a query set. So we consider that all data items in a query set need to be organized as a whole. In this paper, we propose a useful strategy for scheduling broadcast programs in wireless environments. Three contributions are made in this work. First, the problem size can be reduced if we use a coarse-grain unit to organize data. Conventional unit, data item, is too fine to solve the NP-hard problem in less time. So we change the granularity to achieve a better run time. Second, we show that this scheduling can be done by the divide-and-conquer paradigm. This also can help us solve this problem in less time and in a more systematic way. Third, this problem can be solved near-optimally if we map this problem to another similar problem and use its ready-made results. We can avoid exploring redundant properties in this problem domain. Namely, we find out the link between this problem and some existent ones. The rest of this paper is organized as follows. First, we give the symbols used in this paper and formulate the broadcast program scheduling problem. Then, we show the NP14 more pages are available in the full version of this document, which may be purchased using the "Add to Cart" button on the product's webpage: www.igi-global.com/article/transformation-techniquescheduling-broadcast-programs/74168?camid=4v1 This title is available in InfoSci-Journals, InfoSci-Journal Disciplines Computer Science, Security, and Information Technology, InfoSci-Select, InfoSci-Select, InfoSci-Computer Systems and Software Engineering eJournal Collection. Recommend this product to your librarian: www.igi-global.com/e-resources/libraryrecommendation/?id=2