Article Preview
TopIntroduction
Actual trends in mobile telecommunication show rapid growth of Internet related applications, and ever growing demand for them. The phenomenon of convergence in means of communication protocols, services and terminals accelerates this process: mobile applications are able to become more and more popular; users are willing to access Internet resources from their portable devices seamlessly, anytime and anywhere. The continuously growing number of mobile users generates an increasing demand for more widespread and more sophisticated support of mobility, such creating serious challenges for the today’s Internet architecture, which is the TCP/IP stack.
The Internet Protocol (IP) itself was designed in the 1970's, when all hosts of the early Internet were connected using wires: they were fixed hosts, not able to change their network point of attachment. That is why the basics of TCP/IP systems were not designed with any kind of mobility in mind. However, nowadays users are much rarely interconnected by wires: a remarkable mass of modern Internet devices are mobile and thus require the support of frequent changes in their network point of attachment. The shortcomings which make this support hard to provide come from the early days of the Internet. The most important one is the double role of IP addresses. On one hand an IP address identifies the host on the global network: all communication sessions initiated from or terminated at a given terminal is identified by its IP address. On the other hand, IP addresses have a topological locator role: a special network identifier belongs to IP addresses telling the position of the node on the Internet. In other words IP addresses have dual significance (i.e., being identifier and locator at the same time), thus becoming semantically overloaded. These two roles make things complicated and inconvenient when the host starts to move: if the node changes its network point of attachment (and thus its IP address), active communication sessions (which are mostly connected to the TCP/IP numbers) are interrupted and even lost in many cases. Obviously users want ubiquitous connection with seamless handovers and uninterrupted sessions, so engineers started to find an answer here.
One of the solutions for the above introduced problem space is a brand new protocol, which is called Host Identity Protocol (Moskowitz & Nikander, Host Identity Protocol (HIP) Architecture, 2006). HIP is a novel approach which decouples IP addresses from applications by proposing a new, cryptographic namespace to identify hosts or other network entities while IP addresses will remain to act as pure locators. In this architecture a novel layer (called the Host Identity Layer) provides separation of identifier (ID) and locator (Loc) roles of IP addresses (i.e., ID/Loc split): transport level connections are no more bound to IP addresses but to permanent IDs, which remain the same for the lifetime of the host. HIP such provides sophisticated and secure mobility/multihoming support, and creates a powerful toolset as the basis of several advanced mobility management schemes and extensions.
Our goal in this paper is to provide a broad survey of the existing HIP-based mobility management solutions and also to perform extensive simulation-based evaluation of key performance indicators related to handover events. Within this survey and evaluation we focus on four scenarios and extensions of HIP: the basic HIP mobility solution, and one of the earliest micro-mobility, network mobility and proactive distributed solutions (µHIP, HIP-NEMO, UFA-HIP, respectively). All of the three HIP extensions were developed by us in our previous works but this is the first time they are studied in complex simulation models and well-detailed handover scenarios focusing on the most important handover performance indicators.