由於Wireless Local Area Networks (WLANs)具有穩定以及容易布建及價格低廉等優點，而隨著行動設備的普及，具有上網需求的使用者也越來越多，因而出現了Dense WLANs這項技術，意即access point (AP)部署的十分密集。使用如此高密度的是希望能應付尖峰時刻，眾多使用者上網所產生的需求，但由於使用者的生活習慣，導致流量需求變化量極大，在低流量時期，許多AP仍還是維持著開啟的狀態，此時過多的AP所消耗的電量就等於是一種浪費。而在過去對於減少耗電量的方式大多是使用Resource-on-Demand (RoD)的策略，可以根據使用者的需求，來決定需要開啟多少數量的AP。但隨著現今通訊設備的推陳出新，同時也帶給了省電策略新的可能，我們可以將硬體的新能力結合省電策略，達到更大的省電量。在本論文中，我們利用AP Multi User Multi-Input Multi-Output (MU-MIMO)的通訊技術，不僅可依照當前環境的狀態調整AP的開啟與關閉，還可更進一步調整其天線的組態；同時我們也提出了一個新概念：將其他AP當作衡量自己要否需要開啟的感測器，並使用AP的歷史流量資料去計算AP間的Pearson correlation coefficient (皮爾森相關係數) ，取得它們的流量相關性，以決定出下一個需要打開的AP。在我們的實驗模擬中，與其他現有省電方法進行耗電量的評比，結果顯示我們所提出的方法所消耗的電量是少於其他方法的。 Due to cheapness, stability and easy to deploy, Wireless Local Area Networks (WLANs) are becoming the most popular solution. With the popularity of mobile devices, there are more and more users with Internet demand. In order to cope with such high traffic, the term of Dense WLANs is born. It is such dense that there are thousands of APs per square kilometer. We use such high density AP in order to handle the traffic generated in peak hours. However, heavy traffic does not exist all the time. Due to behaviors and habits of users, the traffic demand changes a lots. During low traffic period, all APs still active, the power consumption of additional APs is wasted. The early researches leveraged Resource-on-Demand (RoD) strategy to reduce power consumption, it can turn on different number of AP according to user demand. Due to the progress of the hardware nowadays, it brings the new opportunity to power saving strategy. We can combine novel abilities of hardware into power saving strategy to assist us to obtain more power-saving. In our paper, we leverage Multi-User Multiple-Input Multiple-Output (MU-MIMO) technology of AP, not only can turn on/off of AP but also can furthermore adjust its antenna configuration. We also come up with a concept that other APs serve as their own sensors. In order to decide the next active AP, we use historical traffic data of each AP to calculate Pearson correlation coefficient between APs to acquire their traffic correlation. In our simulation, we compare our RoD strategies with other method in power consumption. The result show that the power consumption of our proposed solution is less than other method.