Netboot is not an anticipation of the claim because it does not have the "preferences image server".
It is dead easy to avoid, though. Just don't "remove the full version of the operating system from the local device when logging off". Note the word "remove"? Do something like invladiate an encryption key that is needed to make the full version work. That is part of the OS's data, not part of the OS itself. Even just by overwiting a small but crucial part of the "full OS", such as a jump table, will get you the result you want, but not infringe the patent.
The thing about patent is to COMPLETELY ignore the abstract, read and interpret the main claims, and if you really do not know what the claims mean, condescend to read the description.
Above all, remember: It's the independent claims that matter -- almost nothing else does!
The US allows an inventor to publish an invention before filing a patent application (although this then blows their chances of getting a valid patent almost anywhere else in the world). This is normally called a grace period. However, if two inventors file for the same invention about the same time, in the US the right to the patent is determined on who is deemed to have made the invention first.
Contrast this with the European registered design system. This also allows a 12-month grace period whereby a design can be published and then validly registered. However, if two independent design applications are filed for the same design, only the first to file has any prospect of getting a valid registration. Several countries allow a grace period for filing patent applications (but NOT the European system) but as far as I am aware, no country other than the USA still uses first-to-invent. The real problem is that the validity of a patent under the first-to-invent régime is hugely dependent on facts that are not public and can only be revealed by discovery in legal action. This encourages litigation and vastly increases costs. The only people that benefit are the patent attorneys (and even though I am one) I believe that this is not good for technology companies or private inventors.
It is a myth that the first-to-invent system protects private inventors who may have delayed filing while getting funding because inventorship date is determined in a legal challenge - it does not just become apparent automatically. If another filed first, and an interference is raised between them in the USPTO, the costs involve spiral out of control, meaning that the richer person gets the patent, irrespective of who invented first.
It is quite a narrow patent that seems to be looking to provide a spoof internet connection to wireless devices when the can connect to the claimed device but there is no available Internet connection.
I have no idea how anyone could possibly infringe this patent, given that the last-but-two and last-but-one clauses of claim 1 are virtually unintelligible. The later independent claims are even stranger and more obscure.
(But as a patent attorney in Europe, I think that the USPTO and US patent law is strangely weird and broken.)