16:55:29 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Welcome everyone. We'll start on time in 5 minutes. 17:15:11 From Navid Nikaein To Everyone: Dear All, please feel free to drop your questions/comments/feedback here in the chat room. The team will answer you directly. 17:16:22 From nitin (Satvikomm, open source Phy layer Engineer) To Everyone: A basic question: The framework allows me to use OAI5G RAN and opens5GS as core n/w. How about if I want to use SD-core as core network. Is it possible? 17:17:43 From Abdulraqeb To Everyone: Is the mobility management part of O-RAN? 17:18:29 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "A basic question: Th..." You may use any external core network that are not configured directly by our management stack. Those vendors mentioned are natively integrated and could be configured and executed with our management stack directly. 17:20:01 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Is the mobility mana..." In terms of handover it is included in the RC service model (we have a demo on it later on). Do you have something more in mind? 17:21:17 From Noe Yungaicela To Everyone: Does FlexRIC support A1, Y1, and O1 interfaces? 17:22:02 From Arno Troch (imec - UAntwerpen) To Everyone: Does the RAN support multi-operator (i.e., multiple PLMNIDs)? 17:23:11 From Sonali Chaudhari To Everyone: Does it support mobility for UE in 3D space? (eg for simulating UAVs) 17:23:15 From Navid Nikaein To Everyone: Replying to "Is the mobility mana..." Mobility management is an important features for a number of use-cases such as QoS gurantee, load balancing, and energy saving among the others. 17:23:58 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Does FlexRIC support..." As of today, a subset of FlexRIC xApps support A1 interface that is used for some more advanced demos later on. The Y1 interface is not support and the configuration of the RIC is done through our own interfaces in the SMO (not through O1 directly). The kernel of the source code is open source and you are welcome to implement any missing features if needed. Regardless, all the features required are implemented in the MX-ORS and MX-PDK products. 17:24:24 From Abdulraqeb To Everyone: Replying to "Is the mobility mana..." Does it have a diffe... 17:24:28 From Navid Nikaein To Everyone: Replying to "A basic question: Th..." You also have the option to ob-board the SD-CORE into the Open RAN studio using our DevOps toolbox 17:25:12 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Does the RAN support..." As of today OAI does not support this but Amarisoft supports multiple PLMNs. Hence, it depends on the vendor. 17:26:23 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Does it support mobi..." As of today, the OAI RF simulator does not implement mobility, but you may connect real UEs to test or use tools such as programmable attenuators to simulate the behavior. 17:26:46 From Navid Nikaein To Everyone: Replying to "Does the RAN support..." Indeed, it depends on the vendor. This is said the BubbleRAN O-RAN stack (i.e. SMO and RIC), multi PLMN/Operators/Slice is fully supported. 17:27:58 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Is the mobility mana..." One major difference to mention is that O-RAN allows for enforcing non-intuitive handovers. For example caused by the energy saving policy rather than just the signal strength. This opens new opportunities to manage the connectivity and mobility beyond the standard 3GPP use cases. 17:28:10 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "You also have the op..." with πŸ‘πŸ» 17:28:22 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "Indeed, it depends o..." with πŸ‘πŸ» 17:28:32 From Bimo (NTUST) To Everyone: what is FB in enc/dec? 17:28:46 From Navid Nikaein To Everyone: Replying to "Does it support mobi..." This is indeed a good addition to the Open RAN Studio. As of now, we support only 2D, but for future use-cases such as those found in ISAC, support of 3D devices are quit important. 17:29:16 From Pablo Picazo (UC3M) To Everyone: Replying to "Does it support mobi..." Regarding this question, I think that you could plug UE emulators like Amarisoft that implement mobility and consider different channel models. 17:29:35 From Praveen (CTU) To Everyone: Can you make sharing more liable by removing the share screen info.. 17:30:09 From ChiehChun Chen (EURECOM) To Everyone: Replying to "what is FB in enc/de..." FlatBuffers https://flatbuffers.dev/ 17:30:19 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "Regarding this quest..." with πŸ‘πŸ» 17:30:28 From Navid Nikaein To Everyone: Replying to "Is the mobility mana..." These are the network-centric handover decisions found in use-cases such as load balancing and energy saving. 17:31:03 From Ali Zaidi To Everyone: Do you have Nephio blueprint for MX PDK yet? 17:31:57 From Navid Nikaein To Everyone: Replying to "what is FB in enc/de..." Additional encoding and decoding for the O-RAN service models. Please have a look at the original FlexRIC paper: https://www.eurecom.fr/publication/6737/download/comsys-publi-6737.pdf 17:32:14 From Pablo Picazo (UC3M) To Everyone: Reacted to Regarding this quest... with "πŸ‘πŸ»" 17:32:16 From Pablo Picazo (UC3M) To Everyone: Removed a πŸ‘πŸ» reaction from "Regarding this quest..." 17:32:28 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Do you have Nephio b..." MX-PDK comes itself as a full package and does not need another SMO or OAM (it has built-in support for Kubernetes Operators). We offer the solution also as a turn-key where a remote installation from OS up to the application could be done. 17:33:25 From Praveen (CTU) To Everyone: Are you running the OAI Core in CU/DU mode for the core part in this demo showing communicating with E2-Agent and XApps. 17:35:05 From Gian Michele, Dell'Aera, TIM To Everyone: Do you think that is needed implement a standard interface between xApp and RIC in order to allow xApp portability? 17:35:50 From Ali Hadian To Everyone: How about Ethernet PDU session? 17:37:01 From Abdulraqeb To Everyone: Replying to "Is the mobility mana..." It means that the ha... 17:37:06 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Are you running the ..." The demo is running monolithic RAN but it can be done also using CU-DU split in the RAN. I am not sure what you mean by the CU-DU split in the core network. Could you please elaborate? 17:37:47 From Navid Nikaein To Everyone: We are showing multiple vendors with the O-RAN stack including OAI and srsRAN with CU-DU split as well as industrial grade Amarisoft and LiteON in monolithic mode. 17:38:06 From Praveen (CTU) To Everyone: Reacted to "We are showing multi..." with πŸ‘ 17:39:43 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Do you think that is..." Your point is valid. In fact FlexRIC uses a mirror of the E2 interface to simplify the interfacing between the xApps and the RIC. This interface is reffered in the literature as E42 or E2*. OSC uses other formats of the interfacing 17:39:44 From Hojjat Navidan (Ghent University/IMEC) To Everyone: As it supports multi-RAT, Is traffic steering between technologies (e.g., vertical handover between LTE/NR) supported? (WG1.Use Case 5) 17:40:45 From nitin (Satvikomm, open source Phy layer Engineer) To Everyone: Reacted to "You also have the op..." with πŸ‘πŸ» 17:40:58 From Navid Nikaein To Everyone: Replying to "Is the mobility mana..." For example gNB load or energy consumption. Have a look at RRM MLB (mobility load balancing) policy. 17:41:53 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "How about Ethernet P..." Ethernet PDU session is supported by Amarisoft only if your UE supports this payload. OAI UE does not provide this feature. Your core network should also support this (we suggest Amarisoft) 17:42:09 From Arno Troch (imec - UAntwerpen) To Everyone: Replying to "Does the RAN suppo..." So how do you test that? Using Amarisoft? 17:42:10 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "For example gNB load..." with πŸ‘πŸ» 17:42:19 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "We are showing multi..." with πŸ‘πŸ» 17:42:55 From Navid Nikaein To Everyone: Replying to "Can you make sharing..." I modified some parameters. Could you check? 17:43:24 From Navid Nikaein To Everyone: Replying to "what is FB in enc/de..." E2AP and E2SM message encoding and decoding based 17:43:38 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "As it supports multi..." Yes. Amarisoft supports Handover across the RATs too. It also supports Spectrum Sharing between RATs 17:44:59 From Navid Nikaein To Everyone: Replying to "Do you think that is..." O-RAN has specified some APIs for this purpose. Having said that, RIC to xApp interface also depends on the RAN functions and therefore the support of those APIs across different vendors are not uniform. 17:45:23 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Does the RAN support..." If the deployment is done using our SMO, it uses the provided slice information to configure the gNBs accordingly. Please refer to our documentation at bubbleran.com/docs or our other demos on our YouTube channel for more information. 17:45:56 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "O-RAN has specified ..." with πŸ‘πŸ» 17:47:09 From Praveen (CTU) To Everyone: Replying to "Does the RAN support..." can we set the same time for multiple vendors and how does it effects the performance at xAPPs? 17:47:47 From Hojjat Navidan (Ghent University/IMEC) To Everyone: Reacted to "Yes. Amarisoft sup..." with πŸ‘ 17:48:53 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Does the RAN support..." If you mean to use different PLMNs on different vendors, yes this case is also supported natively in our SMO. The performance of the xApps is scaled based on the data volume processed (i.e., the number of E2 nodes) not the PLMN count. 17:50:20 From Abdulraqeb To Everyone: Replying to "Is the mobility mana..." I know how to handle... 17:51:04 From Arno Troch (imec - UAntwerpen) To Everyone: Replying to "Does the RAN suppo..." Have you by any chance tried different PLMNs on the same vendor (in a network sharing setup)? 17:51:08 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Is the mobility mana..." The parameters are measured for the gNB not the UEs. Energy consumption of the network is in question here 17:52:25 From Navid Nikaein To Everyone: Replying to "Is the mobility mana..." I was referring to save the energy at the gNB (and not UE) by handing over users to the neighboring cells and shutting down the under-loaded gNBs. Have a look at our MWC demo: https://youtu.be/bR4LBo_4IFc?si=ItgV-dhkRPDXprgo 17:54:35 From Mahdi Darroudi [Neutroon] To Everyone: What is your overview and vision to use ISAC based RU? Any PoC with RIS? 17:54:37 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Does the RAN support..." If I'm not mistaken, I have tried before Amarisoft with two PLMNs (which required two AMFs or CNs). The gNB would broadcast multiple PLMNs. I know for sure it works fine with multiple cells in the same base station and I suppose it works with a single cell scneario too. You may refer to Amarisoft's tech academia to double check 17:55:57 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "What is your overvie..." We have a discussion about the ISAC close to the end of the session using PHY and Sensing service models defined and coined by BubbleRAN 17:56:35 From Mahdi Darroudi [Neutroon] To Everyone: Reacted to "We have a discussi..." with πŸ‘ 17:57:08 From jrmg To Everyone: are slice 1 and slice 3 using different dnn?? 17:57:12 From Navid Nikaein To Everyone: Replying to "What is your overvie..." Essentially, what is needed to perform sensing is the raw I/Q samples across mutli-cell, Multiple Antenna, and Multi-RAN/Frequency. This could be done at the RU or at the DU. For this We defined a new service model to enable it. 17:58:30 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "are slice 1 and slic..." They are connected to separate core networks so implicitly for most intents and purposes yes 17:59:11 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "are slice 1 and slic..." However, the DNN (as the name of the DNN not the PDU session) both are internet DNN 17:59:32 From jrmg To Everyone: Replying to "are slice 1 and slic..." how to diferentiate if they have same sst-sd 17:59:49 From Pablo Picazo (UC3M) To Everyone: Are these UE and RAN emulated or what RU and UE hardware is used in the test? 18:00:00 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "are slice 1 and slic..." The xApp also uses the E2 information which contains the cell metadata 18:00:16 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "are slice 1 and slic..." That's how this slicing is superset to the 3GPP network slicing 18:01:18 From Enrico Guarino [TIM] To Everyone: Reacted to "Are these UE and RAN..." with πŸ‘ 18:01:21 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Are these UE and RAN..." Even though the same setup works for the actual UEs (as seen in other demos too), this particular video is using OAI RF Simulator for OAI and SRS ZMQ simulator for the SRS stack. Both of them emulate the network 18:01:37 From Pablo Picazo (UC3M) To Everyone: Reacted to Even though the same... with "πŸ‘" 18:01:40 From Pranav Charjan To Everyone: Reacted to "Even though the same..." with πŸ‘ 18:02:46 From Arno Troch (imec - UAntwerpen) To Everyone: Reacted to "If I'm not mistake..." with πŸ‘ 18:03:06 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Are these UE and RAN..." Using the simulators for slicing demos comes with the advantage of eleminating the radio effects to improve the accuracy of the measurments for throughput especially 18:03:43 From Navid Nikaein To Everyone: Reacted to "Even though the same..." with πŸ‘ 18:05:01 From Tano Bischoff (Fraunhofer HHI) To Everyone: Reacted to "Even though the sa..." with πŸ‘ 18:06:15 From Tano Bischoff (Fraunhofer HHI) To Everyone: Replying to "Are these UE and R..." A follow-up question to that, has the OAI slicing demo (OAI gNB + OAI core) with slice configuration via RC SM already been demonstrated in an over-the-air setup? 18:06:41 From Ioannis Panitsas To Everyone: Reacted to "A follow-up question..." with πŸ‘ 18:07:54 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Are these UE and RAN..." Yes. It works for over-the-air setups too. We have some customers that use this reguarly with mobile phone and USRP radio. 18:08:09 From Tano Bischoff (Fraunhofer HHI) To Everyone: Reacted to "Yes. It works for ..." with πŸ‘ 18:08:11 From Navid Nikaein To Everyone: Reacted to "Yes. It works for ov..." with πŸ‘ 18:09:32 From Navid Nikaein To Everyone: Replying to "Are these UE and RAN..." This could be extended with Asymmetric uplink and downlink RAN slicing. 18:09:47 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "This could be extend..." with πŸ‘πŸ» 18:15:13 From Pablo Picazo (UC3M) To Everyone: Radio is not interrupted in this procedure? What is the latency introduced when we swap the RB used? 18:16:10 From Pablo Picazo (UC3M) To Everyone: Replying to "Radio is not interru..." Ok looks like 50ms for whole procedure right 18:16:24 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Radio is not interru..." No the raido stays connected. It uses RRC reconfiguration message to modify the BWP configuration in the UE. The latency for the xApp remains the same, but depending on the complexity of the reconfiguration, the RAN might need more time to finish all the messages and procedures with the UE 18:16:42 From Pablo Picazo (UC3M) To Everyone: Reacted to No the raido stays c... with "πŸ‘" 18:17:01 From Navid Nikaein To Everyone: Replying to "Radio is not interru..." On average the RRCConnectionReconfiguration procedures requires between 7-15ms. 18:17:25 From Pablo Picazo (UC3M) To Everyone: Reacted to On average the RRCCo... with "πŸ‘" 18:17:46 From ChiehChun Chen (EURECOM) To Everyone: In here the latency for the BWP configuration is measued from the time xApp sent the control message to xApp recevied the control message acknoledgement, which including the time RAN reconfigured the BWP 18:18:00 From Navid Nikaein To Everyone: Reacted to "In here the latency ..." with πŸ‘ 18:18:09 From Pablo Picazo (UC3M) To Everyone: Reacted to In here the latency ... with "πŸ‘" 18:19:16 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Radio is not interru..." The base station acknowledges the message immediately after the control action is received and validated. The CCC needs to be combined with other SMs or using the monitoring services to verify when the procedure is finished at the RAN level. Hence, the action latency in the xApp is reported without considering the RRC reconfiguration delay 18:24:06 From Nick H To Everyone: Reacted to "A follow-up question..." with πŸ‘ 18:27:39 From Nick H To Everyone: Have you simulated/emulated any 5G NR NTN scenarios with the OpenRAN Studio? 18:29:14 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Replying to "Have you simulated/e..." Not in particular. In their recent release, Amarisoft has added support for the NTN but we haven't yet tested using O-RAN scenarios yet. We would happily invest on development and testing of this matter, if we see the interest in the community and BubbleRAN customers. 18:30:04 From Luis Santos (Ubiwhere) To Everyone: Reacted to "Have you simulated/e..." with πŸ‘ 18:30:40 From Nick H To Everyone: Replying to "Have you simulated/e..." OK - thanks @Alireza 18:30:48 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "OK - thanks @Alireza" with πŸ‘πŸ» 18:42:03 From Bimo (NTUST) To Everyone: Do you deploy the entire LLM system out of O-RAN stack?If so how does the LLM interact/inference the E2? 18:44:28 From Tim (NYCU) To Everyone: Hello Ilias and Prof.Navid, I have a question, say the IBN enhanced with LLM/LAMs, won't the inference if, only if the LAMs utilizing closed models like GPT4o /GPT 3.5 (or other cheaper API), would cause the OpEx increased significantly? 18:44:58 From Ilias Chatzistefanidis (EURECOM) To Everyone: Replying to "Do you deploy the en..." The LLM model could be deployed in different levels of the O-RAN stack. Here we showed two places, the xApp (E42) and the rApp level (R1). We deploy it using the OpenAI APIs 18:47:41 From Bimo (NTUST) To Everyone: Reacted to "The LLM model could ..." with πŸ‘ 18:48:42 From Ilias Chatzistefanidis (EURECOM) To Everyone: Replying to "Hello Ilias and Prof..." Yes actually the LLM model's overhead is a real challenge both for the cost (OpEx) and also for other overhead aspects, such as energy consumption, computations, latency. So, we need to find ways to reduce the LLM's overhead, while maintaining sufficient performance 18:49:22 From Tim (NYCU) To Everyone: Replying to "Hello Ilias and Pr..." Let me correct my question, since you've already answer the previous question using OpenAI APIs, what's the significance if this LAMs deployed in the xApps, or specifically to do Traffic Steering continuously? 18:49:37 From Tim (NYCU) To Everyone: Reacted to "Yes actually the L..." with πŸ‘ 18:53:00 From Ilias Chatzistefanidis (EURECOM) To Everyone: Replying to "Hello Ilias and Prof..." LLMs are a new technology, right? So, the signifficance, let's say, here is to see if LLMs are actually capable of optimizing the network control e.g. for continuous Traffic Steering 18:53:44 From Tim (NYCU) To Everyone: Reacted to "LLMs are a new tec..." with πŸ‘ 18:54:13 From Sarah KASDI (University Paris CitΓ©) To Everyone: Reacted to LLMs are a new techn... with "πŸ‘" 18:59:52 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Usage the static noise video is done to reduce the capabilities of the video encoder to take advatage of compression for example. 19:04:21 From Mirna Awad To Everyone: Hello, thanks for these interesting demos. I have question about previous part. Have you implemented xapps (monitoring, slicing, etc..) in python language? or just in c? I am trying to implement slicing xapps in python using flexric and srsran and it seems complicated or not yet supported by srsran. 19:07:48 From Navid Nikaein To Everyone: We will shortly share all the materials with you. You can also check the materials of the previous episodes. https://bubbleran.com/media/ors-series-ep02/ https://bubbleran.com/media/ors-series-ep01/ 19:08:01 From Nick H To Everyone: Thanks BubbleRAN team! An excellent set of demos as an overview of MX-ORS and scenario simulation. Kudos! 19:08:41 From Alireza MOHAMMADI (BubbleRAN) To Everyone: Reacted to "Thanks BubbleRAN tea..." with πŸ‘πŸ» 19:08:46 From Ilias Chatzistefanidis (EURECOM) To Everyone: Reacted to "Thanks BubbleRAN tea..." with πŸ‘πŸ» 19:08:51 From Razvan-Mihai Ursu To Everyone: Thank you so much for the insightful presentations! 19:09:12 From Derya Turan (Siemens) To Everyone: Thank you so much! 19:09:22 From ChiehChun Chen (EURECOM) To Everyone: Replying to "Hello, thanks for th..." For the current stats, we implemented the xApp in C for the RC SM, for the customized SM - slice SM, which can be written in C, Python and Go 19:09:27 From Jorge Baranda (CTTC) To Everyone: Thanks a lot for this really interesting session 19:09:29 From Pranav Charjan To Everyone: Thank you very much for the presentations and demos! 19:09:37 From David LΓ³pez (CTTC) To Everyone: Thank you very much for such an interesting event! 19:09:37 From Jahangir Khan To Everyone: Thank you so much 19:09:39 From Ingmar To Everyone: Many thanks for your presentations and demos!