Is it possible for a LTE UE to obtain the accurate local or UTC time/date from the L1/L2/L3 of LTE network?
I am sure that a LTE UE can obtain the local or UTC time/date from either MME or L4 servers, e.g., NTP servers. However, as I know, the accuracy of these mechanisms is only about in the unit of seconds or even minutes.
Is there any approach for a LTE UE to obtain/sync its clock at the accuracy of milli-seconds?
Thank you very much,
Shu
I don't think there's any reason why the LTE UE cannot obtain time to within 1 msec, or even less than that, using even NTP. But this would require the NTP server to be close to the LTE base station (BS), or even at the LTE BS, and for that NTP server to have a good time reference locally, such as a local GPS receiver.
If you look up Precision Time Protocol (PTP), described in IEEE 1588, and then NTP, you should see that the same basic mechanism is used to factor out network latency. Why not apply those same techniques to LTE? In short, the time request is time-stamped (even if the local UE clock was not previously set, and is just counting time from power-up), its reception at the time server is time stamped, the response of the time server is time stamped, and reception of the response at the UE is time stamped (again, even if that UE time stamp is no more than "time since powered up"). These time stamps allow the UE to determine the time deltas of packets going upstream from UE to server, and back down from server to UE, from which the UE can estimate the downlink latency (the assumption in both PTP and NTP is that uplink latency is equal to downlink latency). So now, the UE only needs to add the downlink latency to the UTC time reported by the time server, and set its clock.
The only thing limiting NTP accuracy to anything as sloppy as seconds or minutes would be how many intervening nodes exist between the UE and the NTP server. So, just locate that time server as close to the BS as you can, or even at the BS. The precision touted for PTP over IEEE 1588 is extremely tight, within less than 1 usec in an Ethernet network, but note, that only holds true if the edge Ethernet switch is a Stratum 1 time reference.
I don't see why either PTP or NTP cannot be applied to LTE networks, and provide time within very good tolerance limits.
I think there is no L3/IP connection between serving eNB and UEs. One way for UEs to obtain local time is through MME. However, the accuracy of this kind of synchronization is not at the level of milliseconds.
BTW, I agree with you that it doesn't make sense if it is hard for LTE network to provide very accurate local time.
On the other hand, 3G or UMTS networks can broadcasts local time/date information through its synchronization channel(s) or broadcast channel. For example, IS-95 and CDMA2000 can provide local time offset from system time and daylight savings time flag through its FL SYNCH channel.
However, when I checked all 13 LTE SIBs of LTE, which should be broadcasted by a LTE network constantly. Different than UMTS network, I can hardly find any fields of these SIBs indicating local time or time zone information to its UEs.
Okay, but on a more philosophical level, I think of LTE and 4G as moving away from specific layer 2 "channels" or "circuits," as 2G and 3G still use, and more into being a generic packet switched scheme, along the lines of Ethernet or WiFi. So I would expect that 4G service providers can (or should?) feel free to provide easy access to more standardized time protocols. By which I mean, time servers that are topologically close to the UEs, even if not right at the base stations.
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology
- Connectivity