Range-based　volatility　aiming　at　financial　high-frequency　data　has　attracted　more　and　more　attention　for　its　more　accurate　estimation　of　financial　asset＇s　volatility.　The　paper　derives　the　optimal　volatility　estimator　in　the　family　of　realized　range-based　multipower　variation,　according　to　variance　efficiency　with　numerical　simulation　and　it　also　gives　its　weighted　estimator,　according　to　unbiasedness　and　variance　effciency.　Meanwhile,　the　paper　expands　the　realized　GARCH　model　under　the　condition　that　the　realized　GARCH　model　is　combined　with　these　estimators.　The　empirical　analyses　show　that　realized　range-based　quadpower　variation　is　the　optimal　volatility　estimator　in　the　family　of　realized　range-based　multipower　variation,　the　weighted　realized　range-based　quadpower　variation　does　get　rid　of　the　inuence　of　calendar　effect　and　the　expanded　realized　GARCH　models　outperform　traditional　EGARCH　model　in　fit　and　forcasting.