On and off for the past few months I've been practicing biofeedback so that I can increase circulation to my hands and feet on demand. (Biofeedback is about learning to control what would normally be autonomic bodily processes. Some people do it with expensive equipment, but if you have good body awareness like I do you may be able to do some of it without even a thermometer for feedback.) And the cool thing is, it's working! I had a really dramatic experience a week or two ago when I had just gone to bed and wanted to warm up my legs. I relaxed while "pushing" heat towards my legs, and within a minute I started to feel prickling, and then a throbbing sensation over the entire surface of my legs. (It subsided after maybe 10-20 seconds, and then my legs were just warm.) How awesome is that?
It's usually nothing so startling, though. In a neutral-temperature environment, I can get my hands to heat up and turn reddish over the course of about 5 minutes, and I'm still not sure if I can do it at all in a cold environment. (For that, I might need to use a hand thermometer, since my body sense isn't good at distinguishing less-cold from more-cold.) While it would be really cool to get to Wim Hof levels of control over body temperature, my goal is to be able to walk barefoot in snow and ice (or just near-freezing temperatures) -- I'd love to go barefoot outside all year 'round.
I'm still very much a novice at this, but it's easy to find time to practice since I take public transportation and don't have a smartphone to keep me otherwise occupied. However, I don't have much to go on in terms of exercises, and I think I could do better with better training materials. I'm not interested in taking a class (unless it comes with a personal recommendation of high efficacy -- there seems to be a lot of snake oil in this field) but I bet there are some good blog posts or books out there. So far all I have is this guided meditation script, which I memorized the gist of and have been applying to myself. Let me know if you find anything else!
Added 2014-04-04: OK, so I was going to leave this as a response to a comment elsewhere, but it probably belongs in the original post:
My general approach has been to observe my body do something I want it to do and try to associate a specific sensation with it. Examples: "A wave of relaxation spreads outwards from my abdomen", "the flesh of my hands starts to feel overstuffed", "my feet start to feel prickly". That's the first part of the training, and that's where the relaxation exercises come in; when I relax, my extremities get more circulation -- and I can observe the various sensations that accompany it. Then I practice trying to replicate those sensations. I imagine them, I try to will them to happen.
My current model for what's going on is that most of the "sensations" I try to replicate are dead ends, but that some of them are bundled up with, uh, motor patterns -- just like the "feeling" of closing your hand into a fist is not just something you can imagine, but something you can will to happen. An example of a dead-end sensation would be "I feel skin against the back of the tips of my fingers", and a useful one would be "I feel my fingers curling".
That's where the feedback comes in. As I practice trying to replicate each of the sensations, some will work and some won't. When we were babies, trying to learn how to control our bodies, we had pretty direct feedback -- vision, touch, proprioception. We *don't* have a very good sense of arteriole diameter or temperature change! If I can detect which sensations are (literally) effective, I can use the same machinery used to learn to walk. For most people, a digital thermometer with 0.1°F sensitivity is necessary to give this feedback; I seem to have had enough practice observing temperature changes that I didn't need it. (I'll probably start using one anyhow just to speed things up!)
Mind you, this is a big jumble of speculation (with some biological knowledge behind it), personal experience, and things I read on the internet. It worked OK for me, but there's possibly a way more efficient approach and almost certainly a way more correct explanation for what's happening.