Adjacent Beams Resource Sharing to Serve Hot Spots: A Rate-splitting Approach

Abstract

This work addresses the uneven traffic demand scenario in multi-beam satellite systems, in which a hot-spot beam is surrounded by cold beams. After partitioning the hotspot beam in different sectors, resource pulling from cold neighbouring beams is allowed following an aggressive frequency-reuse scheme. As a consequence, the level of the co-channel interference within the hot-spot beam increases. A scheme known as Non-Coherent Rate-Splitting (NCRS) is employed to cope with this interference, based on the exclusive use of magnitude channel state information at the transmitter (CSIT). The receiver complexity is increased with respect to full CSIT precoding schemes, which are considered for benchmarking purposes. Different NCRS strategies are analyzed and compared with several partial and full CSIT schemes. The proposed solution not only shows an improvement with respect to partial CSIT benchmarks, but also displays a competitive performance against full CSIT precoders.* The relative gain G t h is used for theoretical purposes. In an actual satellite system, it would be more practical to use instead

Flexible User Mapping for Radio Resource Assignment in Advanced Satellite Payloads

“…In the previous section, power and/or bandwidth were flexibly allocated to the different beams for a fixed assignment between users and beams. Conventionally, users are served by their dominant beams, even though the additional freedom provided by a flexible mapping between users and beams can be exploited to balance the beam load, as illustrated in previous works dealing with the hot-spot scenario [18], [19], [?]. As in the previous first level designs, the goal is to pose a convex problem for the practical implementation of the resource allocation at beam level.…”

“…We can write g(x) in (19) as the sum of separable functions, so that the optimization problem for the vector x = (x 1 , . .…”

“…with cst a positive constant. Since the offered traffic must be lower than the requested traffic, i.e., log 2 (1 + x • snr b ) ≤ S req (b), we have that the second derivative is negative for the range of admissible values of x and, in consequence, problem (29) and, equivalently, (19), is convex.…”

Non-coherent rate-splitting for multibeam satellite forward link: practical coding and decoding algorithms

“…• Because independent per-antenna coding is sub-optimal, R (NCRS * ) ⊆ R (NCRS). • For Γ = (14,4,4,14) dB, Γ = (10,10,10,10) dB and Γ = (10,10,8,8) dB, R (FDM) ⊆ R (NCRS * ). Here, R (X) ⊆ R (Y) (R (X) R (Y)) indicates that R (X) is (is not) a subset of R (Y).…”

“…S-MAP O max q∈Q Xq O (X ) Table I COMPUTATIONAL COMPLEXITY OF THE RECEIVER PAIR PER SYMBOL PERIOD. 2) A time offset significantly increases the complexity order of both N-MAP and U-MAP. 4 3) In the presence of a time offset, it is not guaranteed that N-MAP yields a lower order of complexity than U-MAP. Everything depends on modulation orders used.…”