Baseband processors

Do you think the Mill architecture is suitable for baseband processor chips? Given the known vulnerabilities of being a public-facing modem, and the fact that verified OSS can’t compete effectively due to the stringent certification requirements, I’d say that the Mill memory model, permission model, and safe stack, would be most welcome to help get it right the first time. You’d probably be able to skip the predictor, since a software radio has to be hard-real-time anyway (thus engineered assuming worst-case everything).