Enabling prosumers to charge their own produced energy at arbitrary locations is a new concept currently being researched. Unfortunately, this concept requires intensive communication between the feed-in system and the electric vehicle (EV). In previous publications, we showed that fluctuation-sensitive model predictive communication is able to reduce the amount of communication necessary by up to 68.2 % while introducing a mean error of 4.11 %. In this poster, we introduce techniques, which reduce the mean error for our system under test to 0.812 %.