From: Mobile and disconnected computing

Mobile-IP [Myles & Perkins, 93] `allows packets between mobile hosts
or networks and other hosts (including fixed hosts) to be delivered
along close to optimal routes'.  Compatibility with existing IP
implementations is one of the key problems in Mobile-IP.  For example,
[Perkins et. al, 93], have suggested a scheme based upon the loose
source routing option of IP packets, but most existing IP
implementations do not implement this option.  Scalability is yet
another important issue.

The Columbia scheme [Ioannidis et al., 91] uses IP-in-IP
encapsulation, thus avoiding problems with non-conforming
implementations; but it achieves only sub-optimal routing for mobility
across widely distributed locations [Aziz, 94].  Some efficient
implementations of IP-in-IP encapsulation capable of supporting
near-optimal wide area mobile routing have been suggested [Aziz, 94],
but more experimentation is required.

For resource-constrained mobile computers, hosting a full IP protocol
suite may be an unacceptable resource burden.  Being able to gateway
with a lightweight protocol to a network node which is hosting a
`heavyweight' protocol suite is a valuable capability [Hildebrand,
94].  Lightweight protocols can also make better use of the bandwidth
limitations of wireless communications.

Apart from this, architectures and implementations that handle the
impact of mobility at higher layers have also been proposed -- such as
the connection-oriented services discussed by Katz [Keeton et. al.,
93], and the mobile socket interface discussed by Casey [Casey, 93].
Current trends would appear to suggest that some form of Mobile-IP
will soon become standard, whereas connection maintenance and caching
in higher-level protocols still needs to be resolved.