Character Trie vs Reverse Word Trie

Evaluating CPU Execution Speed versus Memory Allocation (RAM) when processing a single network request (ads.doubleclick.net). This demonstrates why Nova Shield utilizes a Subdomain Trie architecture.

Character Trie Traditional

CPU Execution Speed 0 / 19 Ops

Iterates through every single character sequentially. High CPU overhead.

Memory Allocation (RAM) ~15 MB

Stores 1-byte character nodes. Highly memory efficient.

Subdomain Trie Nova Shield

CPU Execution Speed 0 / 3 Ops

Splits by domain intervals. Skips entire words per cycle. Ultra-low latency.

Memory Allocation (RAM) ~45 MB

Stores complete String objects in Hash Maps. Marginally higher memory footprint.

The Engineering Verdict: By sacrificing a negligible ~30 MB of modern device RAM, Nova Shield reduces CPU workload by over 84% per network request. In a proxy server handling hundreds of concurrent connections, this trade-off is essential for maintaining zero-latency routing and preserving battery life.