Distributed Ledger Technologies (DLTs) (the agnostic term for blockchain) are a potential solution for many pressing issues arising in the Internet of Things (IoT) domain. These issues include facilitating secure transactions between IoT devices and immutably recording data. Most DLT architectures were not designed with IoT in mind and consequentially do not satisfy the requirements of many IoT applications. However, the relatively new class of Directed Acyclic Graph (DAG) based DLTs show great promise for IoT networks. These DLTs require the rate at which transactions are issued and disseminated to be explicitly managed in order to ensure fairness among users. We present a congestion control algorithm for these DLTs, which optimises dissemination rate and guarantees that all nodes receive the same information and have fair access even in a dishonest environment, subject to the computing limitations of nodes. Our algorithm takes inspiration from well-known areas of networking research, such as QoS, and TCP. However, an important distinction between the DLT setting and traditional networks is the unique nature of traffic in DLT networks and the fact that nodes cannot trust familiar feedback measurements, such as packet acknowledgements or congestion notifications. Our solution realises a decentralised congestion control algorithm for DLTs without the need for trust among nodes.