tbf, I suspect that the main obstacle to that kinda monitoring would be cost and hassle.
*If* a faraday cage / similar fucks up GPS reception, then could someone bork their tag by covering it in foil / some other temporary material? Without proof that it's been intentionally upfucked, you'd be looking at - potentially - a huge number of alerts for reasons of fucking around, as well as proper malfunctions / alerts. If tunnels fuck up GPS, you'd be looking at alerts left, right and center (unless people were specifically excluded from entering tunnels, or the existence of tunnels was coded into the alert software. What about tunnels near schools, or on school routes?)
On top of that, you'd have far, far heavier costs attached to monitoring. atm, you've essentially got one kind of alarm. It is 7pm. Frank is not within x metres of his assigned post. ALARM! -> G4S call the police, or probation, or whatever, and someone might (or might not) get onto it.
With GPS... on top of any costs attached to getting people out to regularly replace / charge GPS tags (could you really ask offenders to charge their own tags?! Again, how long is a realistic battery life for an ankle-sized device?), you've got 24 hour constant monitoring. Not only of curfews, but of danger zones and exclusion zones, as well as other outages / glitches. And tunnels. 24 hour constant monitoring of... how many people? What proportion of alarms, of what kind, and of what duration, do you respond to? What kind of staffing would e.g. G4S need in place to implement that kinda scheme, and what kinda policing decisions / responses would be considered appropriate? Bring it up at a Probation supervision? Or full, emergency police response? Every time you get that call wrong, you're exposed to public attention / scrutiny. So the likelihood is that people low down the food chain will tend to respond by up-tariffing or overestimating risk, leading to reasonably consistent over-responses. (At least, that's what's tended to happen in MAPPA, and in other bureaucratised 'risk monitoring' networks).
Long story short, it *might* just about be feasible / viable for a very small number of offenders who're perceived to be very high risk (or high visibility). But the - IMO unavoidable - costs attendant on anything approaching that kinda technology at the moment would be way, way beyond prohibitive.
To be monitored requires someone to do the monitoring.
Increase the number of people being monitored, and the number of potential places / situations / times at which alarms might be triggered, and you're beginning to get into territory that's even more of a ball-ache for the people organising the monitoring, than it is for those people being monitored.