LTE/WiFi Aggregation (LWA) prospects and its capabilities?

Hi Everyone,

I have been recently following into the LTE/WiFi Aggregation (LWA) topic and need some views and feedbacks from your side. We want to go into trials and see the prospects in Europe.

Why I am following it? No contention problem compared to LTE-U and LAA-LTE, Standard solution compared to MPTCP (Since it requires a GW server in my CORE Network, which is not cheap as well and will be different from country to country and operator to operator)

My questions are:

  1. What are the major points need to be considered that are not mentioned in 3GPP standards before implementing in LTE eNB and WiFi APs?
  2. Does LWA allows me to implement VoWiFi? Or needs WiFi signalling to CORE network? Looking forward to share views.



I got an answer in Linkedin

LWA gives you supplemental capacity to offload non-priority data over WiFi. Split between LTE and WiFi happens at PDCP level in the APs. CN still sees a single S1 connection towards LTE eNB. Signaling and VoLTE calls will be carried over LTE. If you need VoWiFi capabilities with LWA, you’ll have to use non-3GPP access based mechanism defined in the specifications (ePDG etc.)

Now I have question:
For my understanding, so if I want to use VoWiFi, the WiFi APs is connected to ePDG in CN via IPSec Tunnel, that is straight forward process. However, in LWA scenario, I will have only 1 transport from eNB to ePDG, not via IPSec Tunnel, do I need a spliter (which I do not know how the CN will differentiate before ePDG) to split the LTE and WiFi traffic? Or VoWiFi is not possible, then LWA allows only offloading non-priority data, but not voice over Xw interface through an eNB to EPC (S1 Link)


Dear Bikram_Bura,

I’ve been working for almost 3 years on LWA ( we actually proposed something earlier than LWA, that we called Very Tight Coupling between LTE and WiFi). We even developed it and implemented it on a real-time testbed. So I think I can provide some answers.

  1. Major points to consider: It is not clear whether 3GPP intends to use WiFi and LTE simultaneously, or just one at a time, in our case we tried both. So if you use them both simultaneously, I think the most important conclusions we end up with is that if the difference between the delays on WiFi and LTE is too large, the throughput will be worse when using both then when using a single interface. So you need very powerfull traffic steering mechanisms to be sure that will not in this case, and to always make sure to use a “good” WiFi AP.

The 3GPP did not consider at all how to connect WiFi AP to LTE eNodeB. We studied that a lot and it is not so easy as it looks like. For our case, our work is a part of a European project and we consider a converged fixed-mobile network where both accesses share the same layer 2 network. So the connection is easier than when having both networks completely independant.

For VoWIFI, we did not think about it, but I think what I said just now might help. Do not forget that LWA intends to reuse existing WiFi APs and not having new ones only to this purpose. So you may be using loaded APs with a lot of users already associated, and thus more or less bit rate and higher collisions and interference.

For further information on what I just said, we already published couple of papers. Just type Very Tight Coupling between LTE and WiFi. We have some interesting results also.

If you have any other questions, do not hesitate to answer, or contact me directly.

Best regards,


Why would you want to use VoWiFi with LWA to begin with? With LWA, “WiFi is hidden”, so you just use normal VoLTE, which would use seamlessly and transparently LTE, WiFi or both.


Hi DSpider,

Thank you for your reply and expert comments. It is true if the delay between the LTE and WiFi APs is too high and also if the WiFi AP is not good (in terms of traffic/bandwidth and signal strength), the throughput will not be good rather worse.

As per 3GPP LWA Rel 13 standard, it supports LTE and WiFi at the same time. There is an alternative called LWIP based on IP layer tunneling suports only 1 at a time (either LTE or WiFi).

I did read your paper before I saw your comment. It is an interesting read and have understood your test results. Currently we are even in discussion how much latency shall be entertained between LTE eNB and WiFi APs that maintains a legitimate throughput, otherwise it is not worth it (which you also mentioned). It will not support the GBR based services and how to build in the intelligence of routing and splitting of packets.

With best regards,


Dear DSpider,

i am working too on LTE/WIFI Aggregation ( very tight coupling) ; but i have a question are there any research papers about different scheduling techniques concerning the Uplink and downlink transmissions ? any useful journals or transactions if possible

1 Like