This thesis presents a travel time estimation-based task offloading scheme in the large scale of edge computing architecture to improve the intelligent transportation system. In architecture, each base station on the road is equipped with an edge server that can provide short travel time for the requested user within the edge server region. To determine the short travel time, the edge server calculates the road distance from the start point (user position) to the destination point. Due to the centralized manner of data exploitation in the edge server region, the user cannot request short travel time for far destination (i.e., the destination point locates in another edge server). Therefore, a task offloading scheme is adopted to estimate the travel time for the requested edge server for determining the short travel time for a large area. Using the task offloading scheme, we also analyze the delay time of the system, instead of estimating the travel time. Then, the congestion value of each road is considered, which can affect the edge server in determining the short travel time. To predict the congestion value, several methods are used to support the travel estimation-based task offloading. Finally, this proposed system shows the estimated travel and delay time for two schemes and two scenarios through estimation travel time with task offloading and without task offloading. Moreover, we also showed the predicted congestion value and accuracy of each method.