A position-dependent information services, value-based method of data prefetching

Abstract: Prefetching LDD-based strategies such as taking into account data from the DDP, but did not take into account the probability of data access and data update frequency and size of these issues for value-based data prefetching (CDP) strategy, some important data prefetching factors such as access probability, update frequency, data item size, distance and range of data are all included in the value function, the function value based on the size of the value of choice is pre-fetched data. by experiment, CDP strategy is more effective than DDP improve the cache hit rate.

Abstract: LDD-based prefetching strategies like DDP take the data distance into account, but do not take into account the access probability of data, updating data and size of frequency. For these issues, this paper proposes a value-based data prefetching (CDP ) strategy, and some important data prefetching factors, such as access probability, update frequency, data item size, data distance and range of data are included in the value function. We can choose the prefetching data based on the size of function value. By comparing the experiment, CDP is more effective than DDP strategy to improve the cache hit rate.

Keywords: location-related information services, location-related data, data prefetching, the cache hit rate
Keywords:: location-dependent information services, location dependent data, data prefetching, cache hit ratio

0 Introduction
Mobile computing environment, a weak network connection, low bandwidth so users can not obtain the required information in a timely manner, in particular query the location data (Location Dependent Data, LDD, it is easy for the user to change the location of a result of outdated or query results incorrect. The data prefetching can significantly improve data access speed and full use of radio bandwidth [1].

A value-based data prefetching strategies
1.1 The location of the model data position data (LDD, is its value depends on the specific location of the data, LDD has a specific scope of application.

Effective range of the data area (Valid Scope Area, is the effective range of the geometric data instance area. Each LDD has a specific instance of the effective range, and only within this range, the instance is correct.

Data from the (Data Distance, is an instance of MC current location and data the distance between the effective range.

1.2 CDP CDP prefetching methods proposed strategy, prefetching based on the value of the value function to select, pre-fetch value function is as follows: Cost = Puseful × (benefit-penalty (1
Type (1 Puseful the probability for the MC access to LDD, benefit for the MC prefetching LDD gain value, penalty as punishment for prefetching LDD price.

1.2.1 The cost of data prefetching data prefetching incentive to the local cache, not all MC required data is, after computing will become effective after the query data is needs of users, and only this part of the data in order to MC query access to bring benefit. In this paper, fbenefit (di di, said the benefits of data prefetching value function, that MC does not prefetch data access time and data prefetching to reduce the proportion of the access time.

1.2.2 access to the probability of LDD MC access for the possibility of a certain probability of LDD, mainly in the MC after the effective range of the data probability and the probability of future access to the data as the basis, therefore the MC after the effective range of possible future data columns candidate set for the pre-fetched to consider the following two main factors C.: ① from the time perspective. more long-lost data is updated, indicating that the data due to server-side update failure caused by the possibility of pre-fetch data smaller, but The more long-lost data being accessed on its older, again, the less likely it is accessed. ② to think in terms of space. Research shows that the location information service data access, MC move along a certain path through the The higher the probability of the data from the current location closer MC, and the data the greater the effective range of size of area, or move closer to MC the current path or direction of movement of the LDD on the more easily accessible.

1.3 alternative pre-fetch data, prefetch choose to take the data in the MC goal is to the premise of limited resources, making possible the pre-fetched data is MC needs, and as much as possible to provide an effective query information.

Taken in the data selection process should consider the following two situations:
① When S = 0 (the cache is full, regardless of whether C has a surplus of not prefetching LDD, will stop prefetching.

② When 0 <S(缓存还有剩余空间且size(i> S, the MC current location and the cache according to the remaining space should be pre-fetch data to calculate the total size.

Links to Research Papers Download http://www.hi138.com 2 simulation and performance analysis
Experiments to prefetch data in the cache hit rate as the index test compared. Test a set of workload queries randomly generated sequence, composed of 100 queries, each query criteria fields generated, conditions and data sheet values in accordance with certain rules are randomly generated. will MC the cache size is set to the experimental data were 10% of the total, 15%, 20%, 25%, 30%, respectively, in five experiments, the experimental results shown in Figure 1 Fig.

3 Conclusion
In a mobile environment, data prefetching is effective in improving access speed and reduce data access time, a viable approach. In this paper, the possibility of considering the probability of LDD and MC access to every valid query data can provide much information, to design a pre-fetch value choice function, find the candidate focused on prefetching data, as long as these data appear in the broadcast channel to prefetch to the local cache. through the experimental comparison, CDP strategy than DDP, DHP is more effective strategies to improve the cache hit rate.

References:
[1] Li emblem, Yang Bing, Chen Hui, et al. Mobile environment to support real-time transaction processing data prefetching [J]. Computers, 2008,31 (10:1841-1847.

[2] Yin L, Cao G. Adaptive power-aware prefetch in wirelesa networks [J]. IEEE Transactions Wire1ess Communications, 2004.3 (5:1648-1658.

[3] Jiang Z, Kleinrock L. Web prefetching in a mobile environment [J]. IEEE Personal Communications, 1998,5 (5:25-34.

[4] Persone VDN, Grassi V, Morlupi A. Modeling and evaluation of prefetching policies for context-aware information services [C]. Proceedings of the 4th Annual International Conference on Mobile Computing and Networking ,1998:55-65.

[5] Zheng B, Xu J, Lee D L. Cache invalidation and replacement strategies for location-dependent data in mobile environments [J]. IEEE Transactions on Computers, 2002,51 (10:1141-1153. Links to free papers Download Center http://www.hi138.com