The second approach [14] adopted for hardware implementation of shortest-path algorithms employ multiple processing elements (PEs) such that each PE corresponds to a node in the graph. Information about the links is stored in an external memory such that each link is represented using three values namely the index of the start node, the index of the end node and the weight of the link. A completely different graph topology can be supplied to the FPGA-based accelerator as input by changing the contents of the external memory.