Comments on: Can UMTS technology really provide useful WAN connectivity? http://blogs.nil.com/blog/2009/02/10/can-umts-technology-really-provide-useful-wan-connectivity/ The Official NIL Blog Wed, 07 Jul 2010 12:33:01 +0000 hourly 1 http://wordpress.org/?v=3.0.4 By: Stefan Dominko http://blogs.nil.com/blog/2009/02/10/can-umts-technology-really-provide-useful-wan-connectivity/comment-page-1/#comment-57 Stefan Dominko Fri, 27 Mar 2009 13:38:04 +0000 http://blogs.nil.com/?p=545#comment-57 I have noticed behaviour that you described. Of course I have done also tests with several hundred of pings with 100bytes nad 1400bytes packets and got same results as I have already mentioned in response time table. I think that response time primary depends on UMTS provider equipment, oversubscription, signal strength and free resources (cells) on base station in that particular moment. Regarding this, it is a low posibility to have the same response times in different countries or even different regions in same country. I have just done the test to show what one of the UMTS providers is promising and what you really get. Of course in our country (Slovenia). I have noticed behaviour that you described. Of course I have done also tests with several hundred of pings with 100bytes nad 1400bytes packets and got same results as I have already mentioned in response time table. I think that response time primary depends on UMTS provider equipment, oversubscription, signal strength and free resources (cells) on base station in that particular moment. Regarding this, it is a low posibility to have the same response times in different countries or even different regions in same country. I have just done the test to show what one of the UMTS providers is promising and what you really get. Of course in our country (Slovenia).

]]> By: dominico http://blogs.nil.com/blog/2009/02/10/can-umts-technology-really-provide-useful-wan-connectivity/comment-page-1/#comment-53 dominico Fri, 20 Mar 2009 08:51:35 +0000 http://blogs.nil.com/?p=545#comment-53 Try starting one ping with 100 bytes packet size and then after a another parralel ping with 1400 bytes packet size. You should notice decrease of response time for 100 bytes ping. I get less than 70 ms response time for a 100 bytes packet size on HSDPA in a real network. This is because of the radio resource allocation schemes. If you are observing ping response times every five minutes for just a few packets ping that's not the right approach as it takes approx 0.5 seconds for a base station to allocate new resources to terminal. Try starting one ping with 100 bytes packet size and then after a another parralel ping with 1400 bytes packet size. You should notice decrease of response time for 100 bytes ping.

I get less than 70 ms response time for a 100 bytes packet size on HSDPA in a real network.

This is because of the radio resource allocation schemes.

If you are observing ping response times every five minutes for just a few packets ping that’s not the right approach as it takes approx 0.5 seconds for a base station to allocate new resources to terminal.

]]>