Global Connectivity in IP Networks
Sammanfattning: Modern humans like freedom, both in thought and movement. Modern humans are also often
connected to internet. But those connections aren’t as free as the humans themselves. Wireless
makes it easier to obtain this freedom. But in traditionally wireless networks you are only able
to move inside the radio coverage of the access points. I will make an analogy and liken this
to an animal at a zoo. The animal has its cage and will not be able to move outside its borders.
To widen the habitat the cage doors needs to open and a key is required. The key is mobile IP
but the key has its limitation. To move between networks, using mobile IP, it must close the
connection before it can open a new. Just as if the animal had to unlock and lock the door at
every passage.
Good enough?
Might be!
But the modern human like no borders or at least doesn’t want to note them, the handovers
between networks should be seamless. Multi Homed Mobile IP is a solution. It allows you to
have several connections at the same time but you are only using one for traffic. This gives a
possibility to change connections for the traffic without the overhead. It will be, in a way, just
like if the animal was both inside and outside the cage in the same time. Thus the door is open
it can freely move and its habitat been larger.
Now let’s continue the cage analogy but in a different way. We have the animal we have the
cage and the cage has several openings where the animal keeper brings the food. We have
also got one food supply from where the food is brought and several roads from that supply to
the different openings.
In our analogy the food is our network traffic, the cage openings are mobile IP Foreign
Agents through which the traffic are maintained. The food supply is the mobile IP Home
Agent with whom the mobile IP Mobile Host are registered and via all traffic goes and the
roads are the networks between the Home Agent and the Foreign Agents. The fact we’ve got
several cage openings is what makes this Multi Homed Mobile IP.
The animal wants his food as fast as possible and has to decide which opening should be used.
But the only thing he can see is how big these openings are. This is just the way WLAN does
it uses Signal to Noise Ratio which only give the signal strength and nothing else. Thus our
animal doesn’t know anything about how crowded the openings are or the condition on the
roads from the supply. Even if he chooses the biggest hole he can’t be sure to get any food at
all. He needs something else that can tell him about the best route to choose for his food.
The same are for Multi Homed Mobile IP, new metrics are needed.
Christer Åhlund has in his Doctoral Thesis “Extended Mobile IP and support for Global
Connectivity in Hybrid Networks” proposed a Running Variance Metric and a Relative
Network Load as performance metrics. He has also done simulations which show how these
metrics perform.
This Bachelor’s Thesis evaluates these metrics in a real environment and compares the results
against the simulated ones.
The result is that the metrics will serve fine in handover decisions. I’ve also made proposals to
changes in the Running Variance Metric algorithm. These changes are to improve the
performance.
The results has also been used in Robert Brännströms Doctoral Thesis Mobility Management
in Heterogeneous Access Networks
KLICKA HÄR FÖR ATT SE UPPSATSEN I FULLTEXT. (PDF-format)