| 引用本文: | 沈志萍,陈军勇,邬依林.随机性参数分布式量化估计及其最优比特分配[J].控制理论与应用,2016,33(8):1074~1080.[点击复制] |
| SHEN Zhi-ping,CHEN Jun-yong,WU Yi-lin.Distributed quantization estimation and optimal bit allocation for a random variable[J].Control Theory and Technology,2016,33(8):1074~1080.[点击复制] |
|
| 随机性参数分布式量化估计及其最优比特分配 |
| Distributed quantization estimation and optimal bit allocation for a random variable |
| 摘要点击 3150 全文点击 1295 投稿时间：2015-06-01 修订日期：2016-05-16 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2016.50485 |
| 2016,33(8):1074-1080 |
| 中文关键词 最优比特分配 量化信号 最优设计 分布式算子 分布式量化估计 Lloyd-max量化器 最小均方误差 |
| 英文关键词 optimal bit allocation quantization signal optimal design distributed algorithms distributed quantization estimation Lloyd-max quantization minimum mean-square error |
| 基金项目 国家自然科学基金项目(61273109, 60774057), 广东第二师范学院教授博士科研专项经费(2014ARF25), 广东省科技计划项目(2014A090906010, 2016A010106007), 河南师范大学博士科研启动经费(5101019170158), 河南省高等学校重点科研项目(16A120005)资助. |
|
| 中文摘要 |
| 本文研究总比特率给定下随机向量参数分布式量化估计及其最优比特分配问题. 与现有文献大都假定每个传
感器的量化比特率给定而不是最优分配下研究随机性参数的分布式量化估计问题不同的是, 本文将综合考虑最优量化
器、最优估计器算法以及给定总比特率下的最优比特分配问题. 针对向量状态标量观测模型, 首先借助现有文献给出基
于量化观测的最优估计器及其误差协方差阵形式表达, 其次得到各传感器的渐近最优量化器实际为著名的
Lloyd-max量化器, 且各传感器的渐近最优量化级数与信噪比成正比, 同时引入一种次优的求解非负整数比特率的方法.
考虑到当传感器数目比较大时, 初始的最优估计器算法运算量很大, 设计了一种渐近等价的迭代量化估计器算法, 其计
算负担大大减轻, 且对于存在延迟或丢包的网络环境亦适用, 增强了算法的鲁棒性. 仿真结果表明, 本文提出的最优比
特分配方案估计性能明显优于一般的均匀比特分配方案. |
| 英文摘要 |
| This paper studies distributed quantization estimation and optimal bit allocation problems of a random vector
parameter given a total bit rate. Different from the existing literature generally assumed that each sensor quantization bit
rate is given rather than optimal bit allocation in researching the corresponding problem, this paper will combine the design
of the optimal quantizer, the optimal estimator algorithm and the optimal bit allocation problem under a given total bit. For
a vector state scalar observation of an observation model, we first give the optimal estimator and its error covariance matrix
in form based on the quantitative observation with the existing literature, and then to get a conclusion that the asymptotic
optimal quantizer of each sensor is actual the famous Lloyd-max quantizer, and that the asymptotic optimal quantitative
level of each sensor is proportional to the signal-to-noise ratio (SNR), at the same time, we introduce a suboptimal method
of solving the non-negative integer bit rate. Considering when the number of sensors is larger, the original optimal estimator
algorithm computational complexity is very big, we design a asymptotic equivalence iterative quantization estimator
algorithm, which can greatly reduce the calculation burden, and can apply to the network environment with some delay or
packet loss, so this method can also enhance the robustness of the algorithm. Simulation results show that our designed
method can achieve a significant amount of the estimation MSE reduction when compared with the uniform allocation
scheme in which each sensor quantizes its observation with the identical bit. |
|
|
|
|
|