Wireless networks analysis

We need to provide a results during AIS/CAD

What is done:

• debugged ns-2.33 code and resolved a multitude of problems in interaction between 3G module and the module providing multi-interface capability;
• debugged ns-2.33 code and resolved problems with TCP working improperly in /some/ simulations;
• added the auto-testing script;
• added calculation of mean delays and signaling costs to calculating script;
• looked at different ns-2 enhancement packages, and found ns-miracle package. Its benefits: * utilizes dynamic loading of libraries in script (works in ns-2.33 and later) * has nice implementation of WLAN by default * has flexible node architecture (borrowed from an early version of Q2S, which we use).
  Though the ns-miracle package didn't prove to be useful for me, it might be useful for other people in the lab doing network modeling. (Short presentation with samples from the distribution maybe? But only if I have some free time in August, which is unfortunately very unlikely.)

Results:

• The 3G model now works OK
• Both UDP and TCP agents work OK with 802.11 and PMIPv6 modules.
• We have two independent (and working) models: 3g-handover and wlan-handover. 3g model covers only TCP, though UDP can be easily added (1.5-2 hours MAX with testing).

What should be done (in order of importance):

• any other QoS parameters, except delay and bandwidth and PMIPv6 signaling cost. Which parameters? Do we need them? DISCUSSION NEEDED. It's reasonable to start with this --- it is the simplest task to do.
• 3g-wlan handover according to 3GPP TS23.402 and others. For now I don't have a good idea how to implement it, I need to study technical specification 23.402 more thoroughly, and then DISCUSS. The main points of discussion will be "what (inessential) things can be cut from the procedures to make it more simple to implement and study".
• traffic tags. if we want to send control traffic thru 3g, and data traffic through WLAN, we need to tag what is what. Fortunately, ns-2 already has such thing (flow-identifiers), but we need to put them to use (classification of packets). Also need to DISCUSS possible bandwidth for control and data traffic. Could we measure it instead of guessing?

What is impossible to do [in less than two months of hard work]:

• IPSec, AAA server and B-type 3G handover (which uses both of them). IPSec is the most hard part, maybe it could be skipped (emulated by delay, ..., whatever.) THIS NEEDS DISCUSSION

Where I am now: I am currently on vacation, and will be available soon (after 05.08, I think).

Current problems: * Time estimates given are overly optimistic. Lots of work at NIC is likely to arrive when I return from vacation (which ends on 03.08). More realistic estimate is "half-done 3g-wlan handover model with some new QoS parameters measured @ the beginning of September". Though it probably need not be wholly for the presentation to be written :-))) * I am very tired, and two weeks of rest is unlikely to solve this. But we will see :-)

Possible future problems: * NIC is a time hog. We will be rushing to finish the half-baked components of Matematika-PIK in september, october, and november (and, realistically speaking, december too) :-(