You are managing a streaming pipeline using Dataflow with hopping windows to group incoming data. However, some late data is not being recognized as such, causing inaccuracies in downstream aggregations. What measures should you implement to correctly capture and process this late data within its intended window?