- Probabilistic flooding: Relay probability adapts to network size (30-100%)
- Bloom filter duplicate detection: O(1) lookups with 4096-bit filter
- Connection pooling: Reuse connections with exponential backoff
- Adaptive TTL: Reduces hops based on network size (2-5)
- Message aggregation framework: 100ms batching window
- BLE advertisements: Include network size/battery hints
These optimizations improve capacity from ~20-30 to ~50-100 users while
maintaining privacy (no routing tables) and simplicity. The system now
adapts automatically to network conditions without configuration.
Trade-offs: Slightly higher CPU for bloom filter, probabilistic relay
may miss edge cases, but overall much better scaling behavior.
