The report describes models that have been developed for sensor fusion and communication in a sensor network. They can be used for modeling and simulations of the effects of sensor fusion on a system level. They are designed for being a part of a large simulation environment that includes massive real time simulations and dynamic user interactions. The models will be implemented as a module in the simulation framework Naval Operational Analysis (NOA), where they will be used for real time simulations. The models focus on the common factors that are essential for the performance of a sensor network; synchronization, delays and network topology. They can handle various types of sensor data provided that the analysis of the sensor data is expressed as a state estimate (for example position or velocity) and a covariance describing the uncertainty in the state estimate. A description of the uncertainty in the sensor data estimates is necessary to be able to perform sensor fusion. The possibilities for sensor fusion is limited by the network's logical and physical topology. The logical topology describes how data can be transported within the network and it is more or less independent of the physical topology. The physical topology focuses on the hardware and describes how sensors, crypto, routers, data analysis centrals etc. are physically connected. In the report both perspectives on sensor networks are described. For the logical topology, both centralized and various kinds of decentralized fusion are discussed. Aspects on the physical topology that clearly affects the sensor fusion are described; data types, amounts of data, data rates and protocols for data links. The software framework of the communication and sensor fusion models is described. Centralized and decentralized sensor fusion can be modelled, with focus on the communication perspective. Results of the computations will vary depending on time delays and available bandwidth. Insufficient bandwidth, in relation to the data that is transmitted, will also create time delays. With the framework it is possible to identify some of the bottlenecks in a sensor fusion network.