Before configuring the MIPv6 testbed, you will need to make some policy decisions:

• What kind of global network prefixes you are going to use in your network? 6BONE prefixes (3ffe::/16) are now deprecated so you will need to obtain a suitable production prefix from the 2001::/16 address block by contacting your ISP.
• How are you going to implement the IP routing? In a small network ripng would be a good choice. In both, Linux and in FreeBSD, you can use zebra (http://www.zebra.org) for accomplishing this.

Since the MIPv6 specification is relatively new (only recently received RFC status) and is still somewhat work in progress, implementations tend to be staggered in terms of what features they support. Thus, interoperability between implementations is somewhat of a hit and miss affair. Therefore, it is wise to first deploy the same MIPv6 implementation on all the machines in the testbed (so make sure that the implementation has HA, MN and CN functionality). Once confidence in one implementation is achieved, you may decide to deploy other implementations and see if they are interoperable.

Note that movement detection algorithms rely on receiving router advertisements from the default IPv6 router when first entering a network. Intelligent movement detection algorithms will also make use of media disconnect/connect notifications and issue router solicitation messages to speed up the reception of these router advertisement messages. Therefore, it is essential that the routers pertaining to the networks in the MIPv6 testbed issue periodic router advertisements so that a MN can configure itself with a new Care of Address when it connects to the network.

The following sections will try to throw some light on how a MIPv6 testbed can be accomplished by using open source software like FreeBSD and Linux in addition to commercial software such as Microsoft Windows XP/CE and Cisco IOS.