Networks
Pure networking concepts — RAN architecture, 5G NR, scheduling, beamforming, latency budgets, slicing. Each node carries a mental model, mathematical model, failure modes, and an AI×Networks connection across 4 rounds.
Complete — 6 nodes, Round 4 ✓
O-RAN Architecture Round 4 ✓
CU/DU/RU functional split · near-RT & non-RT RIC · xApp lifecycle. The hub node: the substrate that makes AI-in-the-RAN deployable.
5G NR Numerology & Frame Structure Round 4 ✓
Scalable subcarrier spacing, slots, and the 2⁻ᵘ ms clock — the hard real-time deadline every scheduler and ML control loop must beat.
RAN Scheduling Algorithms Round 4 ✓
Round robin, max C/I, and proportional fair (sum-log utility) — the per-slot decision that is the cleanest RL target in the RAN.
Fronthaul Constraints & Latency Budgets Round 4 ✓
eCPRI, the ~100 µs HARQ budget, CPRI vs 7.2x split — the two numbers that set the functional split and the latency gradient.
Beamforming & Massive MIMO Round 4 ✓
Array gain, spatial multiplexing, zero-forcing, favorable propagation — and why CSI aging makes it the richest ML target in the PHY.
Network Slicing + SLA Guarantees Round 4 ✓
Isolation vs efficiency, statistical multiplexing, SLA-as-constraint — the canonical target for constrained RL resource allocation.