Exploiting Collaborative Computing to Improve Downlink Sum Rate in Satellite Integrated Terrestrial Networks

In this paper, we target at improving the downlink user sum rate in satellite integrated terrestrial network (SITN). Wherein, the densely deployed low earth orbit (LEO) satellites provide backhaul to terrestrial base stations (TBSs) and TBSs serve the ground users (GUs) over different bands. Note that the user sum rate is dependent on the user association and backhaul capacity. Moreover, the backhaul capacity is determined by multi-connectivity feature between the TBSs and LEO satellites. However, the densification and high dynamics of SITN aggravate the inter-satellite interference, which degrades the backhaul capacity. To this end, we propose a joint interference management and user association (JIMUA) scheme, which exploits collaborative computing to enhance the backhaul capacity and satisfy the user demands. The service demands of GUs and dwell duration of LEO satellites can be captured by introducing a collaborative computing coefficient. With this aid, the TBSs can predict the highly dynamic trajectory of dense deployed LEO satellites. Accordingly, the searching space of satellites can be reduced, the interference can be mitigated and backhaul capacity can be improved with high time efficiency. Moreover, the user association is achieved by dividing GUs into different groups, while the required user demands can be satisfied. Considering the dense deployment of LEO satellites, the user sum rate of SITN could be enhanced by JIMUA compared with the benchmark. With the well tuned collaborative computing coefficient, the backhaul capacity can be increased over 12%. Consequently, the user sum rate can be improved over 7%. Moreover, with the increasing user density, the gap between the user sum rate of JIMUA and the optimal solution can be reduced to less than 4%.