A caching strategy to reduce network impacts of PCS
A caching strategy to reduce network impacts of PCS
复制标题
减少 PCS 网络影响的缓存策略
DOI：
10.1109/49.329333
复制
发表时间：
1994
期刊：
影响因子：
--
通讯作者：
S. Mohan
中科院分区：
文献类型：
--
作者：
R. Jain;Yi;C. Lo;S. Mohan
我们提出了一个辅助策略，称为每用户缓存，定位用户谁从一个地方移动到另一个地方，同时使用个人通信服务（PCS）。高速缓存策略增强了在诸如GSM和IS-41的现有标准中提出的基本定位策略，其目的是减少网络信令和数据库负载，以换取增加的CPU处理和存储器成本。由于技术趋势正在推动后者的成本下降，辅助战略将变得越来越有吸引力。缓存背后的想法是重用有关被叫用户位置的信息，以便随后呼叫该用户，并且对于那些经常接收呼叫的用户（相对于他们更改注册区域的速率）非常有用。这种想法试图利用用户接收的调用的空间和时间局部性，类似于利用计算机系统中文件访问的局部性的想法。我们使用一个参考PCS架构和用户的本地呼叫移动性比（LCMR）的概念来量化使用缓存和用户类的成本和效益，这将是有益的。我们还提出了两个简单的算法来估计用户的LCMR和情况下，每一个是优选的。我们表明，在各种假设下，缓存可能会产生显着的净效益，减少信令网络流量和数据库负载。>
We propose an auxiliary strategy, called per-user caching, for locating users who move from place to place while using Personal Communications Services (PCS). The caching strategy augments the basic location strategy proposed in existing standards such as GSM and IS-41, with the objective of reducing network signaling and database loads in exchange for increased CPU processing and memory costs. Since technology trends are driving the latter costs down, the auxiliary strategy will become increasingly attractive. The idea behind caching is to reuse the information about a called user's location for subsequent calls to that user, and is useful for those users who receive calls frequently relative to the rate at which they change registration areas. This idea attempts to exploit the spatial and temporal locality in calls received by users, similar to the idea of exploiting locality of file access in computer systems. We use a reference PCS architecture and the notion of a user's local call-to-mobility ratio (LCMR) to quantify the costs and benefits of using caching and classes of users for whom it would be beneficial. We also present two simple algorithms for estimating users' LCMR and the situation in which each is preferable. We show that under a variety of assumptions caching is likely to yield significant net benefits in terms of reduced signaling network traffic and database loads. >