He was pretty protective in his opinion in Kyllo (the whole "general public use" assertion being deemed a potential factor, not a determination, in the restriction of an expectation of privacy), though it was centralized on the place as holding an expectation of privacy, not the person.

Though he dissented in Bond (2000), the combination of those two rulings may give fairly powerful protections against technological surveillance.

I was thinking of the decidedly non-technological search in Hudson v. Michigan actually, though I take your point that his record is mixed.