Abstract : Web-based applications are increasingly demanding many computationally intensive services. On the other hand, FPGA-based hardware accelerators(HwAcc) provide good performance in accelerating computationally intensive applications. In addition, some FPGAs support a dynamic partial reconfig-uration (DPR) techniques to virtualize and share the FPGA underlying hardware resources in time multiplexing during run-time to save resource and power consumption. Integrating FPGA in a cloud environment is an indispensable way to improve efficiency and provide acceleration services to demanding users. More importantly, in recent years it was proved that FPGA resources deployed in a cloud environment can be accessed with the same OpenStack software technology used to access virtual machines. However, the performance of the virtualized FPGA is highly dependent on the communication medium used to interconnect the virtualized FPGA resources and the control manager. After analyzing the possible interconnect mediums, we have selected Network-on-Chip (NoC) which support parallel communication as the efficient medium for accelerators. Consequently , we propose a NoC based virtualized FPGA as cloud Services. Two virtualized FPGA-based cloud service: Hardware Accelerator as a Service(HAaaS) and Reconfigurable Region as a Service(RRaaS) are proposed in this paper. The NoC provides layered and parallel communication between the virtualized regions of the FPGA and helps them to communicate their status and exchange data through the routers connected to them. A 2x2-mesh NoC based reconfigurable accelerators for image analysis and matrix computation are implemented and tested showing a promising result for more scalable systems in cloud computing.