The use of low-orbit constellation of small or super-small satellites for solving problems of remote sensing of the Earth is a promising direction for the development of space activities. In order to ensure the communication of these satellites with ground-based points in advanced space systems, it is planned to use groupings of low-orbit communication satellites. At that, a certain formation of this grouping is considered. It is assumed that the satellites are evenly distributed in several orbital planes, and each of them has a connection with two neighboring satellites in its plane and two satellites in neighboring planes. The object of research in the article is the problem of routing, namely, the search for routes for relaying data streams from remote sensing satellites to ground-based information reception points. The proposed approach to solving the routing problem uses the following network specifics. The nodes of the network are remote sensing satellites, communication satellites and objects of ground infrastructure. In this case, you can highlight two fragments of the network. Communication satellites form the first fragment of the network, and the second one — the communication channels of these satellites with ground infrastructure objects and remote sensing satellites. The topology of the first network fragment is static, and the topology of the second fragment is dynamically changing. However, the dynamics of changes in the topology of this network fragment is predictable. It can be calculated based on satellite flight simulations and described as a contact plan that defines the time parameters of satellite communication sessions with ground stations. The solution of the problem is based on an agent-based approach. Satellite agents form the overlay layer of the network and, based on information interaction, provide route search, traffic balancing, and data transmission without delays at network nodes. The article offers information interaction schemes that provide both centralized and distributed route search options.