Out of curiosity, does anyone know, according to the LTE spec, where in 
particular the down/up AMBRs are actually being enforced on the network?  Is it 
all taking place at the EPC?  Or does the EPC inform the eNB of UE-AMBR and 
have the eNB take it into account as it schedules things?  Or is the UE 
informed of it when it attaches to the network and it is the UE's 
responsibility to enforce it in whole (both down/up) or in part (just up)?

When I first dived into DOCSIS, I remember being shocked to learn that the 
bandwidth constraints were values included in the provisioning file that was 
fed to the cable modem, and it was in fact up to the modem to enforce the 
specified throughput limits.  I think the explanation I remember hearing is 
that it is better to have the limit enforcement distributed instead of 
centralized because at least in the uplink direction, the modem would be free 
to schedule/queue packets it was transmitting in an intelligent way rather than 
force the network to throttle things more primitively by simply throwing 
packets out (which will slow down TCP, but not necessarily contain a barrage of 
UDP).  However, this meant that vulnerabilities in modem firmware could allow 
for circumvention/abuse by the customer (which is not just academic and has in 
fact happened).  So there are some trade-offs.  I am interested to know which 
way the 3GPP leaned.  (And perhaps the scheduling mechanisms that LTE uses make
  the centralized vs. distributed debate a non-starter anyway.)


Nathan Anderson
First Step Internet, LLC

Telrad mailing list

Reply via email to