Abstract: With the explosion in the number of Internet of things (IoT) and connected vehicle networks in smart city, the challenges to meet the demands from both data traffic delivery and data computing are increasingly prominent, and the allocation of network resources has attracted great attention. A novel network architecture based on mobile edge computing (MEC) was proposed in this paper to incorporate connected vehicle networks and IoT networks to transmit the delay-tolerant data and execute the computing tasks. In order to integrate diverse and complex standards and protocols in the same network, the programmable control principle originated from software-define networking (SDN) paradigm was introduced. Moreover, the process of delay-tolerant data transmission and computing node selection in software-defined vehicle network was formulated as a partially observable Markov decision process (POMDP) to minimize the system cost, which consists of both the network overhead and execution time of computing tasks. Simulation results show that the system cost can be decreased efficiently compared with the existing schemes, the processing time of computing tasks can be shorten and the computing efficiency can be improved. Furthermore, the arrival rate of delay-tolerant data can also be ensured within the delay requirements.