To meet current legislation limits, modern diesel engines already achieve very low raw emission levels and utilize additional components for aftertreatment. However, during fast transients still undesired emission peaks can occur for both soot and NOx. These are caused by differences in the in-cylinder conditions between the quasi steady state engine calibration and the transient engine operation, e.g. during tip-ins. These effects become more and more important in view of future RDE emission test cycles. In this work a case study is performed to analyze the potential reduction of transient soot emissions during a specified engine maneuver. An additional target is to investigate potential benefits of a novel in-situ soot sensor based on the Laser Induced Incandescence (LII) principle which offers a high temporal resolution. The transient soot emissions and the sensitivity to selected injection parameters, such as timing or rail pressure, are investigated by experiments at a passenger car diesel engine where the soot emissions are measured by two different sensors, a production standard and the novel prototype soot sensor. The novel LII sensor is mounted directly in the exhaust manifold downstream of the turbocharger and provides additional insights in the transient behavior. The measurements of production type and prototype sensors are used to develop time varying models which will be utilized in a numerical optimization problem and to determine an optimized injection profile. The modeling and optimization task is carried out in parallel for both sensor devices and the resulting trajectories are applied to a common rail passenger car diesel engine. In the experimental validation the optimized injection profiles, obtained from both models, allow to reduce the undesired overshoots during transients with minimal impact on the torque response. The novel sensor allows to gain further insights in the relation between input parameters and soot response, which is reflected by the obtained models. These models, applied in optimization, then lead to a better transient result than the models derived from the production standard soot sensor measurements.