The cognitive network controller (CNC) defines a learning agent that can adapt online and in near real-Time the routing decisions needed for bundle transmissions in a space delay-Tolerant network and other challenged networks. The agent uses a spiking neural network (SNN) as the learning element in a reinforcement learning loop, which incrementally optimizes the outbound link selection for each bundle based on its estimated routing cost. In this paper, a digital hardware implementation of the SNN element of the CNC is proposed, which helps to accelerate the routing decision-making process to faster-Than-real-Time levels. The design was tested on a Zynq Z7020 (PYNQ-Z2) SoC/FPGA board. A distributed implementation of the CNC is also proposed, which allows offloading the SNN execution from a DTN gateway to a remote device that hosts the FPGA implementation of the SNN. The methods were validated using an emulated satellite network testbed.