Elsevier

Information Sciences

Volume 242, 1 September 2013, Pages 76-91
Information Sciences

Broadcasting on-demand data with time constraints using multiple channels in wireless broadcast environments

https://doi.org/10.1016/j.ins.2013.04.026Get rights and content

Abstract

Data Broadcasting is an effective approach to provide information to a large group of clients in ubiquitous environments. How to generate the data broadcast schedule to make the clients’ average waiting time as short as possible is an important issue. In particular, when the data access pattern is dynamic and data have time constraints, such as traffic and stock information, scheduling the broadcast for such data to fulfill the requests is challenging. Since the content of the broadcast is dynamic and the request deadlines should be met, such data broadcasting is referred to as on-demand data broadcasting with time constraints. Many papers have discussed this type of data broadcasting with a single broadcast channel. In this paper, we investigate how to schedule the on-demand broadcast for the data with time constraints using multiple broadcast channels and provide two heuristics to schedule the data broadcast. The objective of the proposed heuristics is to minimize the miss rate (i.e., ratio of the number of requests missing deadlines to the number of all requests) and latency (i.e., time between issuing and termination of the request). We show that the offline version of the considered problem is NP-hard and present a competitive analysis on the proposed heuristics. More discussion about the proposed heuristics is given through extensive simulation experiments. The experimental results validate that the proposed heuristics achieve the objectives.

Introduction

Advanced technologies in wireless communications, information systems, and hand-held devices make it possible for mobile clients to conveniently access different kinds of information services, such as electronic news, traffic information, and stock prices. In such an environment, the bandwidth between a server and a client is asymmetric [1], [8], that is, the downlink bandwidth is much greater than the uplink bandwidth. When the group of mobile clients is large, the bandwidth of the uplink becomes a bottleneck in traditional client–server model. In wireless environments the number of mobile clients is growing constantly and the information service system is expected to serve an increasingly large number of users. Thus, how to solve the bottleneck problem becomes a critical issue.

In recent years, data broadcasting has been considered an attractive solution for the bottleneck problem and it also provides an efficient way to disseminate the information to a large pool of clients. In general, data broadcasting can be classified into two types [16], [21]: push-based data broadcast and on-demand data broadcast.

  • push-based data broadcast: In push-based data broadcast, a server periodically broadcasts data items on a broadcast channel. Fig. 1a illustrates the push-based data broadcast, where the server repeatedly broadcasts data items 1, 2, 3, and, 4 and the clients tune into the channel to access the data items. Most of the scheduling algorithms in this type consider static data access patterns and sequences.

  • on-demand data broadcast: In on-demand data broadcast, the clients send the requests via an uplink channel and then the server broadcasts the requested data items. Fig. 1b shows the on-demand data broadcast, where the dashed and solid lines represent the uplink and broadcast channels respectively. The on-demand data broadcast can be used for dynamic and large-scale data dissemination.

In some information services, data may have temporality, such as traffic information and stock quotes, so the data may become invalid as time passes. To ensure that the data is timely, the clients usually request the data with deadlines and the server then broadcasts the on-demand data before the deadlines. The requested data become invalid when the deadlines are passed. Thus, how to schedule on-demand data broadcasts for data with time constraints is an important topic. To our knowledge, although many papers have discussed this topic, most of them considered a single broadcast channel [3], [4], [5], [7], [12], [13], [19], [20], [21]. For a server, using multiple channels [2], [6], [15], [16], [17], [18], [23], [24] to provide information makes the broadcast cycle shorter than using one channel, thus serving more users in a short time.

In this paper, we discuss the problem of scheduling on-demand broadcasts for data with time constraints in multi-channel environments. A broadcast consist of a sequence of broadcast slots (or packets). For simplicity, we assume that each data item corresponds to one slot in the broadcast. We consider that each request has its own deadline and has multiple data items associated with it. Data items are related because of the associated requests. If we lose any one of the data items associated with a specific request, the data items received may become useless. For instance, a client may request a Web page which contains some components, such as audio clips or images. When the client uses a browser to access the Web page, the browser will request all the components contained in the Web page after receiving the Web page. Another application would be when a mobile client has an inquiry about the stock prices of some companies. It is highly likely that the mobile client would like to compare the stock information and make a decision on investment, so the quotes of these stocks therefore are related because of the request. In addition, some location-based services, like range query and kNN query [9], usually have more than one data item associated with the query.

With multiple broadcast channels, the data overlap problem [6] will occur when the requested data items for a request appear in different channels at the same time, and only one channel can be tuned into at a given time instance. More details about the data overlap problem will be given in the next section. The data overlap problem will force the client to wait until the following broadcast cycles to receive the relevant data items, thus leading to a longer latency. However, a data-overlap free broadcast schedule also makes the broadcast cycle long, with many empty slots in the broadcast. In order to reduce the latency, data replication can be used to provide more data items earlier to serve more clients by placing data items in the empty slots. The objective of this paper is to provide on-demand data broadcast schedules where more requests meet their deadlines and reduce the waiting time for the clients. In our design, we avoid the data overlap problem as much as possible and use data replication implicitly to make the broadcast compact and the latency short.

The on-demand data broadcast has following property. The server has no knowledge about all the requests in advance and makes decisions only with the information of the requests already received. This property is similar to the conditions for online algorithms [3]. On the contrary, an offline algorithm is given the whole problem data from the beginning. The performance of an online algorithm can be measured using competitive analysis. An online algorithm is c-competitive if the optimal solution generated by the offline algorithm is c times better than the solution generated by the online algorithm with the same problem instance.

The rest of this paper is organized as follows. After giving preliminaries and related research in Sections 2 Preliminaries, 3 Related work, respectively, we describe the system model in Section 4. The formal definition of the problem and a demonstration that the offline version of the problem is NP-complete are given in Section 5. We then propose two heuristics, MPHH and MPLH, in Section 6. The competitive analysis of proposed heuristics is discussed in Section 7. The simulated experimental results are presented and discussed in Sections 8 Experiments, 9 Conclusions concludes this paper.

Section snippets

Preliminaries

Two measures are usually considered when using data broadcasting to provide data items: latency and tuning time. In this section, we first introduce these two measures. Then we give an another measure, miss rate, which is introduced additionally to evaluate the service quality for the requests having deadlines. Last, we present what the data overlap problem is.

Related work

In this section, we introduce some related works about data broadcasting. One of the objectives of most papers about data broadcasting is to achieve a short latency [1], [8], [14], [20]. To reduce the latency, the proposed broadcast schedules considered the data access probabilities and/or dependency. Furthermore, some researchers use multiple broadcast channels to have more deduction on the latency [2], [6], [15], [18], [23], [24]. In general, the algorithms to generate a broadcast schedule

System model

The on-demand data broadcasting system we considered is shown in Fig. 4, where a large group of clients can retrieve the data items maintained by a server. On the system, the clients send the requests to the server via an uplink channel. Each request requests some data items and has its deadline. Each requested data item has a unique identifier. The server receives the requests and inserts them into a request queue. When the request queue is not empty, the server schedules the data broadcast

Problems

Suppose that there are c channels and n data items, d1, d2,  , dn, in data set D. Each data item is of the same size and takes 1 time slot to be broadcast. On the client side, we assume that there are m requests. Each request Qi, 1  i  m, has a deadline dli and is associated with a set of qi data items, {di(1),di(2),,di(qi)}, where di(j)  D, 1  j  qi  n. We denote the total number of requests for a data item dl as Ndl, 1  l  n. Set Ui indicates the set of unserved data items in the request Qi, 1  i  m,

Proposed heuristics

As mentioned in Section 3, almost all the related papers considered urgency and productivity in order to have better performance in terms of miss rate and latency. The fairness is generally not considered because avoiding the starvation problem may increase the miss rate and latency of the generated broadcast schedule. We follow these and present two heuristics to generate the broadcast schedules based on the urgency and productivity of data items when the number of channels is limited. The

Competitive analysis

In this section, we present a competitive analysis for our proposed heuristics, MPFH and MPLH. We use OPT to denote the optimal solution for the offline version of OBMM problem in the analysis. For simplicity the number of served requests is considered as the measurement instead of the miss rate. In fact, the miss rate and the number of served requests have the same meaning for the measurement. To derive the competitive ratio, we compare our algorithms with OPT in terms of the number of served

Experiments

This section presents the results of the simulation experiments and our findings. We compare our proposed algorithms with three other algorithms, EDF, MSF, and DUP. The DUP algorithm has good performance for on-demand data broadcasting in multi-channel environments without considering the deadlines of requests. Recall that EDF and MSF are the scheduling algorithms for a single broadcast channel. In order to use these two algorithms on multiple broadcast channels, we modify them with the

Conclusions

In this paper, we discuss how to generate a broadcast schedule for on-demand data broadcast with time constraint on multiple channels. The problem is formulated and referred to as the On-demand Broadcasting with Minimum Miss rate (OBMM) Problem. This is an online problem and we show that the offline version of the OBMM problem is NP-Hard. Two heuristics are provided for the problem: algorithm MPFH considers the popular data items first and algorithm MPLH tries to serve more requests by

References (25)

  • J. Chen, V.C.-S. Lee, E. Chan, Scheduling real-time multi-item requests in wireless on-demand broadcast networks, in:...
  • S.-Y. Fu, C.-M. Liu, Broadcast schedules and query processing for K nearest neighbors search on multi-dimensional index...
  • Cited by (0)

    View full text