Source linked

MeshDNS يضرب 0.47ms على ESP8266 مع إزالة الفشل البيزنطية

يكتسب نطاق DNS التعاوني للاتصالات IoT المحددة من الموارد حلولاً تصل إلى مليار ثانية، وتحدد الأخطاء البيزنطية باستخدام التصويت المشترك المقبول بـ Ed25519 على ميكروزرولر 2 دولار.

meshdnsesp8266iotdnsbyzantine fault toleranceed25519

MeshDNS drops warm-cache DNS resolution to 0.47 milliseconds on commodity ESP8266 hardware—roughly a third of native mDNS's 1.39ms—while layering Ed25519-signed quorum voting to isolate Byzantine faults. That's not a theoretical simulation on a workstation; that's real silicon with sub-50 KB usable RAM and an 80 MHz clock.

Sub-millisecond DNS on a $2 Microcontroller

The paper's authors built MeshDNS for exactly the kind of network where a single DNS authority is a single point of failure: resource-constrained IoT deployments with unreliable links and adversarial peers. Each node maintains cache awareness using hash-based summaries, so the whole mesh knows who has what without flooding. Warm-cache queries resolve in under half a millisecond because the node just grabs the record locally; no round trips. That 0.47 ms number comes from an actual implementation on ESP8266—chips you can buy for a couple of bucks.

Trading Latency for Byzantine Safety on Cold Misses

Cold-cache misses are where the clever engineering shows up. Instead of trusting a single responder, MeshDNS requires identical answers from multiple admitted peers, each signed with Ed25519 signatures. The penalty is predictable: about 1.3 to 1.7 seconds per cold miss. That's a deliberate tradeoff—spend a couple of seconds to guarantee Byzantine fault isolation among peers, assuming physical hardware extraction is out of scope. In an environment where a compromised node could poison a whole network's name resolution, that 1.7 seconds is cheap insurance.

Scaling from 5 to 1,000 Nodes Without Central Authority

The team validated MeshDNS on a 5-node physical testbed and then ran discrete-event simulations scaling to 1,000 nodes. The results show resilient local name caches even under churn—nodes joining, leaving, or going silent. Because the framework uses shared-key admission and hash-based cache synopses, it doesn't need a central DNS resolver or a cloud gateway. Every node is both client and resolver.

What this enables next: persistent edge telemetry in sensor swarms where you can't trust every node but you can't afford a full-time DNS server either. MeshDNS is available now on GitHub under the artifact link, so anyone with a few ESP8266s and a soldering iron can start testing Byzantine-tolerant name resolution today.


Source: MeshDNS: A Cooperative DNS Resolution Framework for Resource-Constrained IoT Networks
Domain: arxiv.org

Read original source ->

External source stays available while the OJO article and comment thread stay local.

Comments load interactively on the live page.