State the design axis that everything turns on: in phased-array antennas, you trade between agility, performance, and cost, and where you sit on that triangle reveals your market. A consumer satellite terminal optimizes cost above all. A defense system optimizes agility and performance, because the mission, not the bill of materials, is the constraint.

Northrop's grant US10892549B1 (inventor Ronald P. Smith), classified squarely in H01Q 3/34, electronic beam scanning, sits at the defense corner of that triangle. The single-inventor, single-CPC profile of the filing suggests a focused architectural claim rather than a sprawling systems patent.

“A phased-array antenna system includes antenna elements of an RF front-end that each propagate a wireless beam portion. A digital beamforming system generates a digital beam corresponding to the wireless beam that is transmitted or received from the phased-array antenna system.”— U.S. Patent No. 10,892,549 source

The independent claim's real subject is how the beam is formed, and the architecture it protects is hierarchical digital beamforming. Between the RF front-end and the beamforming system the claim inserts a digital signal conditioner system made up of multiple digital beamforming processors, each tied to a proper subset of the antenna elements, never the whole array. Those processors are collectively configured to process the digital beam in a plurality of iteration levels: a lowest level associated with the individual wireless beam portions at each antenna element, and a highest level associated with the full digital beam. The defining rule is recursive, every digital beam portion at a given iteration level is a sum of lesser, time-delayed digital beam portions from the next lower level.

That structure is what lets a defense-grade array scale. Instead of one monolithic processor digesting every element's signal at once, the beam is built up in stages, with each processor combining a handful of time-delayed sub-beams and passing the partial sum upward. Time delay, not just phase shift, is applied at every level, which is exactly what wideband electronic beam steering needs to keep a beam pointed across a large instantaneous bandwidth. The hierarchy distributes the combining load and makes the element count expandable without redesigning the central processor.

The contrast with SpaceX's contemporaneous uni-dimensional steering work (US11239553B2) is the whole lesson. SpaceX deliberately halved its steering dimensionality to crush terminal cost; the defense filing assumes you want full two-axis electronic agility, true time delay, and a beamforming pipeline engineered to deliver it. Same CPC class, opposite optimization.

Why this matters to anyone tracking the sector: phased-array IP is where the satellite-comms and defense worlds increasingly overlap, and reading the CPC plus the assignee tells you which set of incentives produced a given claim. A claim built around iterative, time-delayed digital combining across element subsets reads as performance-first; the same H01Q 3/34 class on a consumer filing would be built around cost-first shortcuts. Beamforming is the battleground, and the same physics serves a missile-defense radar and a broadband dish very differently.

The dependent claims fill in how the hierarchy is physically partitioned, and the picture that emerges is a binary-tree-like combining structure. Each digital beam portion at a given level is associated with a contiguous group of antenna elements, and that group grows in size from the lowest iteration level to the highest, so the array is summed up in expanding contiguous neighborhoods rather than arbitrary scatterings of elements. Another claim makes the parallelism concrete: a first digital beamforming processor handles the partial sum for one contiguous group of elements while a second processor handles the partial sum for an adjacent group, and their outputs are themselves combined at a higher level. The combining work is spread across many small processors operating on local neighborhoods, with each higher tier merging the tiers below it.

A further claim allows a single processor to span levels, processing both a lowest-level digital beam portion and a higher-level one, which lets the architecture be mapped onto real hardware flexibly rather than demanding one dedicated processor per tier. Read together, these dependents describe a beamforming pipeline that scales by adding contiguous subarrays and the processors that serve them, each new tier doubling down on the same time-delayed summation rule. That is precisely the property a defense buyer wants: an aperture that can grow to thousands of elements, as the description contemplates, without the central beamformer becoming the bottleneck, and that applies true time delay at every stage so a large array can hold a beam across wide bandwidth without the beam-squint that afflicts phase-only steering. The cost-first consumer arrays in the same CPC class make the opposite bargain, trading exactly this kind of distributed, time-delayed processing for cheaper, simpler combining.

The technical framing in the disclosure reinforces which side of the triangle this sits on. Background discussion situates the invention against ordinary phased arrays in which each element propagates a portion of the beam with an associated time delay and amplitude, and for a received beam the element portions are combined and digitized for processing. The claimed advance is to make that combination iterative and digital throughout, so that the beam is reconstructed in software from the bottom up rather than summed in analog hardware at a single stage. That choice trades silicon and processing for flexibility: a digitally combined array can form and steer multiple independent beams from the same elements, adapt its weighting on the fly, and place nulls toward interferers, all capabilities a defense radar prizes and a cost-minimized terminal can do without. The recursive, time-delayed summation rule is the mathematical backbone that lets all of that ride on one expandable architecture.

The limit, as ever, is that a phased-array patent is an architecture, not a fielded radar. It tells you how Northrop intended to steer a beam, that it chose a hierarchical digital-beamforming pipeline, and what it considered worth protecting; demonstrated performance, sidelobe levels, scan loss, and real bandwidth, lives on a test range, not in a claim set.