I was diffing two PCAPs the other day trying to work out what’s up, and noticed the Instance ID on a GTPv2 IE was different between the working and failing examples.
If more than one grouped information elements of the same type, but for a different purpose are sent with a message, these IEs shall have different Instance values.
So if we’ve got two IEs of the same IE type (As we often do; F-TEIDs with IE Type 87 may have multiple instances in the same message each with different F-TEID interface types), then we differentiate between them by Instance ID.
The only exception to this rule is where we’ve got the same data, so if you’ve got one IE with the exact same values and purpose that exists twice inside the message.
Ask anyone in the industry and they’ll tell you that GTPv2-C (aka GTP-C) uses port 2123, and they’re right, kinda.
Per TS 129.274 the Create Session Request should be sent to port 2123, but the source port can be any port:
The UDP Source Port for a GTPv2 Initial message is a locally allocated port number at the sending GTP entity.
So this means that while the Destination Port is 2123, the source port is not always 2123.
So what about a response to this? Our Create Session Response must go where?
Create Session request coming from 166.x.y.z from a random port 36225 Going to the PGW on 172.x.y.z port 2123
The response goes to the same port the request came on, so for the example above, as the source port was 36225, the Create Session Response must be sent to port 36225.
Because:
The UDP Destination Port value of a GTPv2 Triggered message and for a Triggered Reply message shall be the value of the UDP Source Port of the corresponding message to which this GTPv2 entity is replying, except in the case of the SGSN pool scenario.
But that’s where the association ends.
So if our PGW wants to send a Create Bearer Request to the SGW, that’s an initial message, so must go to port 2123, even if the Create Session Request came from a random different port.
What is included in the Charging Rule on Gx ultimately turns into a Create Bearer Request on GTPv2-C.
But the mapping it’s always obvious, today I got stuck on the difference between a Single remote port type, and a single local port type, thinking that the Packet Filter Direction in the TFT controlled this – It doesn’t – It’s controlled by the order of your Traffic Flow Template rule.
Input TFT:
"permit out 17 from any 50000 to any"
Leads to Packet filter component type identifier: Single remote port type
Whereas a TFT of:
permit out 17 from any to any 50000
Leads to Packet filter component type identifier: Single local port type (64)
I’ve written about Milenage and SIM based security in the past on this blog, and the component that prevents replay attacks in cellular network authentication is the Sequence Number (Aka SQN) stored on the SIM.
Think of the SQN as an incrementing odometer of authentication vectors. Odometers can go forward, but never backwards. So if a challenge comes in with an SQN behind the odometer (a lower number), it’s no good.
Why the SQN is important for Milenage Security
Every time the SIM authenticates it ticks up the SQN value, and when authenticating it checks the challenge from the network doesn’t have an SQN that’s behind (lower than) the SQN on the SIM.
Let’s take a practical example of this:
The HSS in the network has SQN for the SIM as 8232, and generates an authentication challenge vector for the SIM which includes the SQN of 8232. The SIM receives this challenge, and makes sure that the SQN in the SIM, is equal to or less than 8232. If the authentication passes, the new SQN stored in the SIM is equal to 8232 + 1, as that’s the next valid SQN we’d be expecting, and the HSS incriments the counters it has in the same way.
By constantly increasing the SQN and not allowing it to go backwards, means that even if we pre-generated a valid authentication vector for the SIM, it’d only be valid for as long as the SQN hasn’t been authenticated on the SIM by another authentication request.
Imagine for example that I get sneaky access to an operator’s HSS/AuC, I could get it to generate a stack of authentication challenges that I could use for my nefarious moustache-twirling purposes whenever I wanted.
This attack would work, but this all comes crumbling down if the SIM was to attach to the real network after I’ve generated my stack of authentication challenges.
If the SQN on the SIM passes where it was when the vectors were generated, those vectors would become unusable.
It’s worth pointing out, that it’s not just evil purposes that lead your SQN to get out of Sync; this happens when you’ve got subscriber data split across multiple HSSes for example, and there’s a mechanism to securely catch the HSS’s SQN counter up with the SQN counter in the SIM, without exposing any secrets, but it just ticks the HSS’s SQN up – It never rolls back the SQN in the SIM.
The Flaw – Draining the Pool
The Authentication Information Request is used by a cellular network to authenticate a subscriber, and the Authentication Information Answer is sent back by the HSS containing the challenges (vectors).
When we send this request, we can specify how many authentication challenges (vectors) we want the HSS to generate for us, so how many vectors can you generate?
TS 129 272 says the Number-of-Requested-Vectors AVP is an Unsigned32, which gives us a possible pool of 4,294,967,295 combinations. This means it would be legal / valid to send an Authentication Information Request asking for 4.2 billion vectors.
It’s worth noting that that won’t give us the whole pool.
Sequence numbers (SQN) shall have a length of 48 bits.
TS 133 102
While the SQN in the SIM is 48 bits, that gives us a maximum number of values before we “tick over” the odometer of 281,474,976,710,656.
If we were to send 65,536 Authentication-Information-Requests asking for 4,294,967,295 a piece, we’d have got enough vectors to serve the sub for life.
Except the standard allows for an unlimited number of vectors to be requested, this would allow us to “drain the pool” from an HSS to allow every combination of SQN to be captured, to provide a high degree of certainty that the SQN provided to a SIM is far enough ahead of the current SQN that the SIM does not reject the challenges.
Can we do this?
Our lab has access to HSSes from several major vendors of HSS.
Out of the gate, the Oracle HSS does not allow more than 32 vectors to be requested at the same time, so props to them, but the same is not true of the others, all from major HSS vendors (I won’t name them publicly here).
For the other 3 HSSes we tried from big vendors, all eventually timed out when asking for 4.2 billion vectors (don’t know why that would be *shrug*) from these HSSes, it didn’t get rejected.
This is a lab so monitoring isn’t great but I did see a CPU spike on at least one of the HSSes which suggests maybe it was actually trying to generate this.
Of course, we’ve got PyHSS, the greatest open source HSS out there, and how did this handle the request?
Well, being standards compliant, it did what it was asked – I tested with 1024 vectors I’ll admit, on my little laptop it did take a while. But lo, it worked, spewing forth 1024 vectors to use.
So with that working, I tried with 4,294,967,295…
And I waited. And waited.
And after pegging my CPU for a good while, I had to get back to real life work, and killed the request on the HSS.
In part there’s the fact that PyHSS writes back to a database for each time the SQN is incremented, which is costly in terms of resources, but also that generating Milenage vectors in LTE is doing some pretty heavy cryptographic lifting.
The Risk
Dumping a complete set of vectors with every possible SQN would allow an attacker to spoof base stations, and the subscriber would attach without issue.
Historically this has been very difficult to do for LTE, due to the mutual network authentication, however this would be bypassed in this scenario.
The UE would try for a resync if the SQN is too far forward, which mitigates this somewhat.
Cryptographically, I don’t know enough about the Milenage auth to know if a complete set of possible vectors would widen the attack surface to try and learn something about the keys.
Mitigations / Protections
So how can operators protect ourselves against this kind of attack?
Different commercial HSS vendors handle this differently, Oracle limits this to 32 vectors, and that’s what I’ve updated PyHSS to do, but another big HSS vendor (who I won’t publicly shame) accepts the full 4294967295 vectors, and it crashes that thread, or at least times it out after a period.
If you’ve got a decent Diameter Routing Agent in place you can set your DRA to check to see if someone is using this exploit against your network, and to rewrite the number of requested vectors to a lower number, alert you, or drop the request entirely.
Having common OP keys is dumb, and I advocate to all our operator customers to use OP keys that are unique to each SIM, and use the OPc key derived anyway. This means if one SIM spilled it’s keys, the blast doesn’t extend beyond that card.
In the long term, it’d be good to see 3GPP limit the practical size of the Number-of-Requested-Vectors AVP.
2G/3G Impact
Full disclosure – I don’t really work with 2G/3G stacks much these days, and have not tested this.
MAP is generally pretty bandwidth constrained, and to transfer 280 billion vectors might raise some eyebrows, burn out some STPs and take a long time…
But our “Send Authentication Info” message functions much the same as the Authentication Information Request in Diameter, 3GPP TS 29.002 shows we can set the number of vectors we want:
5GC Vulnerability
This only impacts LTE and 5G NSA subscribers.
TS 29.509 outlines the schema for the Nausf reference point, used for requesting vectors, and there is no option to request multiple vectors.
Summary
If you’ve got baddies with access to your HSS / HLR, you’ve got some problems.
But, with enough time, your pool could get drained for one subscriber at a time.
This isn’t going to get the master OP Key or plaintext Ki values, but this could potentially weaken the Milenage security of your system.
If you’re working with the larger SIM vendors, there’s a good chance they key material they send you won’t actually contain the raw Ki values for each card – If it fell into the wrong hands you’d be in big trouble.
Instead, what is more likely is that the SIM vendor shares the Ki generated when mixed with a transport key – So what you receive is not the plaintext version of the Ki data, but rather a ciphered version of it.
But as long as you and the SIM vendor have agreed on the ciphering to use, an the secret to protect it with beforehand, you can read the data as needed.
This is a tricky topic to broach, as transport key implementation, is not covered by the 3GPP, instead it’s a quasi-standard, that is commonly used by SIM vendors and HSS vendors alike – the A4 / K4 Transport Encryption Algorithm.
It’s made up of a few components:
K2 is our plaintext key data (Ki or OP)
K4 is the secret key used to cipher the Ki value.
K7 is the algorithm used (Usually AES128 or AES256).
It’s important when defining your electrical profile and the reuqired parameters, to make sure the operator, HSS vendor and SIM vendor are all on the same page regarding if transport keys will be used, what the cipher used will be, and the keys for each batch of SIMs.
Here’s an example from a Huawei HSS with SIMs from G&D:
We’re using AES128, and any SIMs produced by G&D for this batch will use that transport key (transport key ID 1 in the HSS), so when adding new SIMs we’ll need to specify what transport key to use.
In our last post we covered the basics of NB-IoT Non-IP Data Deliver (NIDD), and if that acronym soup wasn’t enough for you, we’re going to take a deep dive into the flows for attaching, sending, receiving and closing a NIDD session.
The attach for NIDD is very similar to the standard attach for wideband LTE, except the MME establishes a connection on the T6a Diameter interface toward the SCEF, to indicate the sub is online and available.
The NIDD Attach
The SCEF is now able to send/receive NIDD traffic from the subscriber on the T6a interface, but in reality developers don’t / won’t interact with Diameter, so the SCEF exposes the T8 API that developers can interact with to access an abstraction layer to interact with the SCEF, and then through onto the UE.
If you’re wondering what the status of Open Source SCEF implementations are, then you may have already guessed we’re working on one! PyHSS should have support for NB-IoT SCEF features in the future.
NB-IoT provides support for Non-IP Data Delivery (NIDD) over 3GPP Networks, but to handle this, some new network elements are introduced, in a home network scenario that’s the SCEF and the SCF/AS.
On the 3GPP side the SCEF it communicates to the MME via the T6a Interface, which is based upon Diameter.
On the side towards our IoT Service Consumers (in the standards referred to as “SCS/AS” or “Service Capabilities Server Application Servers” (catchy names as always), via the RESTful HTTP based T8 interface.
The start of the S1 Attach procedure is very similar to a regular S1 attach.
The initial S1 PDU Connectivity Request indicates in the ESM Message Container that the PDN Type is Non IP.
S1 PDU Connectivity Request from attach procedure
Other than that, the initial attach procedure looks very similar to the regular S1 attach procedure.
On the S6a interface the Update Location Request from the MME to the HSS indicates that this is an EUTRAN-NB-IoT Radio Access Type.
And the Update Location Answer APN Configuration contains some additional AVPs on the APN to indicate that the APN supports Non-IP-PDN-Type and that the SCEF is used for Data Delivery.
The SCEF-ID (Diameter Host) and SCEF-Realm (Diameter realm) to serve this user is also specified in the APN Configuration in the Update Location Answer.
This is how our MME determines where to send the T6a traffic.
With this, the MME sends a Connection Management Request (CMR) towards the SCEF specified in the SCEF-ID returned by the HSS.
The Connection Management Request / Response
The MME now sends a Diameter T6a Connection Management Request to the SCEF in the Update Location Answer,
In it we have a Session-Id, which continues for the life of our NIDD session, the service-selection which contains our APN (In our case “non-ip”) and the User-Identifier AVP which contains the MSISDN and/or IMSI of the subscriber.
To accept this, the SCEF sends back a Connection-Management-Answer to confirm we’re all good to go:
At this point our SCEF now knows about the subscriber who’s just attached to our network, and correlates it with the APN and the session-ID.
On the S1 side the connection is confirmed and we’re ready to roll.
Mobile Originated Data Request / Response
When the UE wants to send NIDD it’s carried in NAS messaging, so we see an Uplink NAS transport from the UE and inside the NAS payload itself is our HEX data.
Our MME grabs this out and sends it in the form of of a Mobile-Originated-Data-Request (MODR) to the SCEF, along with the same Session-ID that was setup earlier:
At this stage our Non-IP Data is exposed over the T8 RESTful API, which we won’t cover in this post.
Note: I’m lazily posting this as its been in my drafts folder for an exceedingly long time – Before going too much further, it’s worth pointing out that eMBMS never really made it anywhere – no production networks of note use eMBMS. I started researching it and my interest petered out once I discovered I couldn’t get any UEs or hardware that supported eMBMS.
Mobile networks are designed as point to point, all traffic is unicast.
But multicast and broadcast traffic is real, and becoming more common in some applications.
In areas where users stream the same radio program, or TV show, live, each of them is consuming the same data stream, but each one gets sent a unique copy of the data, on a resource block allocated to them for reception of the data.
If we have 10 users on a cell, each streaming a 5Mbps live video, that’s 50Mbps of capacity taken up on the radio / air interface. If that stream was moved onto a eMBMS service, only 5Mbps of capacity would be used, regardless of how many people on the cell are consuming it.
For Mission Critical Push to Talk applications, the lack of broadcast/multicast support was highlighted again. For a PTT app with 10 users in a talk group, you’d need to schedule resource blocks for 10 users, and allocate 10 radio resources 10 times, send GTP packets 10 times, all to send the same data to 10 people.
So enter eMBMS – The Evolved Multimedia Broadcast and Multicast Service, providing multicast service for LTE.
Overall Architecture
eMBMS introduces a few changes to the RAN side to handle support for a shared data channel, which is sent by the eNodeB and that UEs can listen on to get data. (More on admission control later)
From a core perspective two new network elements are introduced, the Broadcast/Multicast Service Center (BM-SC) and Multimedia Broadcast Multicast Services Gateway (MBMS GW), these elements function in much the same was the P-GW and S-GW retrospectively, but in regards to Multicast services.
Like so many 3GPP specs before it, MBMS relies on GTP for transporting the data to be distributed, and relies on GTPv2-C for control plane data.
BM-SC – Broadcast Media Service Centre
The Broadcast Multicast Service Centre acts as the gateway between content providers (providing streams of data to be distributed) and the EPC.
The BM-SC sets up eMBMS sessions and pulls broadcast data from the content providers and collects receipts from subscribers of some streams to charge / track consumption of the services.
In this regard the BM-SC is akin to the P-GW, which as the border for the EPC and external networks, except it’s largely unidirectional.
MBMS Gateway
The MBMS Gateway (MBMS-GW) encapsulates the broadcast data stream from the BM-SC and encapsulates it into GTP packets to be distributed to eNBs across the network.
The MBMS-GW allocates a multicast transport address for each broadcast data stream?
MME Interaction
For this a new interface is introduced on the MME – the Sm interface, which interconnects the MME and the MBMS-Gateways assigned to it.
S8 Home Routing is a really simple concept, the traffic goes from the SGW in the visited PLMN to the PGW in the home PLMN, so the PCRF, OCS/OFCS, IMS, IP Addresses, etc, etc, are all in the home network, and this avoids huge amounts of complexity.
But in order for this to work, the visited network MME needs to find the PGW of the home network, and with over 700 roaming networks in commercial use, each one with potentially hundreds of unique APNs each routing to a different PGW, this is a tricky proposition.
If you’ve configured your PGW peers statically on your MME, that’s fine, but it doesn’t scale very well – And if you add an MVNO who wants their own PGW for serving their APN, well you’ll be adding some complexity there to, so what to do?
Well, the answer is DNS.
By taking the APN to be served, the home PLMN and the interface type desired, with some funky DNS queries, our MME can determine which PGW should be selected for a request.
Let’s take a look, for a UE from MNC XXX MCC YYY roaming into our network, trying to access the “IMS” APN.
Our MME knows the network code of the roaming subscriber from the IMSI is MNC XXX, MCC YYY, and that the UE is requesting the IMS APN.
So our MME crafts a DNS request for the NAPTR query for ims.apn.epc.mncXXX.mccYYY.3gppnetwork.org:
Because the domain is epc.mncXXX.mccYYY.3gppnetwork.org it’s routed to the authoritative DNS server in the home network, which sends back the response:
We’ve got a few peers to pick from, so we need to filter this list of Answers to only those that are relevant to us.
First we filter by the Service tag, whihc for each listed peer shows what services that peer supports.
But since we’re looking for S8, we need to find a peer who’s “Service” tag string contains:
x-3gpp-pgw:x-s8-gtp
We’re looking for two bits of info here, the presence of x-3gpp-pgw in the Service to indicate that this peer is a PGW and x-s8-gtp to indicate that this peer supports the S8 interface.
A service string like this:
x-3gpp-pgw:x-s5-gtp
Would be excluded as it only supports S5 not S8 (Even though they are largely the same interface, S8 is used in roaming).
It’s also not uncommon to see both services indicated as supported, in which case that peer could be selected too:
x-3gpp-pgw:x-s5-gtp:x-s8-gtp
(The answers in the screenshot include :x-gp which means the PGWs advertised are also co-located with a GGSN)
So with our answers whittled down to only those that meet our needs, we next use the Order and the Preference to pick our best candidate, this is the same as regular DNS selection logic.
From our candidate, we’ve also got the Regex Replacement, which allows our original DNS request to be re-written, which allows us to point at a single peer.
In our answer, we see the original request ims.apn.epc.mncXXX.mccYYY.3gppnetwork.org is to be re-written to topon.lb1.pgw01.epc.mncXXX.mccYYY.3gppnetwork.org.
This is the FQDN of the PGW we should use.
Now we know the FQND we should use, we just do an A-Record lookup (Or AAAA record lookup if it is IPv6) for that peer we are targeting, to turn that FQDN into an IP address we can use.
And then in comes the response:
So now our MME knows the IP of the PGW, it can craft a Create Session request where the F-TEID for the S8 interface has the PGW IP set on it that we selected.
For more info on this TS 129.303 (Domain Name System Procedures) is the definitive doc, but the GSMA’s IR.88 “LTE and EPC Roaming Guidelines” provides a handy reference.
How does one encode / interpret the value of this AVP / IE was the question I set out to answer.
TS 29.274 says:
For the encoding of this information element see 3GPP TS 32.298
TS 32.298 says:
The functional requirements for the Charging Characteristics as well as the profile and behaviour bits are further defined in normative Annex A of TS 32.251
TS 32.251 Annex A says:
The Charging Characteristics parameter consists of a string of 16 bits designated as Behaviours (B), freely defined by Operators, as shown in TS 32.298 [51]. Each bit corresponds to a specific charging behaviour which is defined on a per operator basis, configured within the PCN and pointed when bit is set to “1” value.
After a few circular references I found this is imported from 32.298.
Finally we find some solid answers hidden away in TS 132 215, under the Charging Characteristics Profile index.
Charging Characteristics consists of a string of 16 bits designated as Profile (P) and Behaviour (B), shown in Figure 4. The first four bits (P) shall be used to select different charging trigger profiles, where each profile consists of the following trigger sets:
S-CDR: activate/deactivate CDRs, time limit, volume limit, maximum number of charging conditions, tariff times;
G-CDR: same as SGSN, plus maximum number of SGSN changes;
M-CDR: activate/deactivate CDRs, time limit, and maximum number of mobility changes;
SMS-MO-CDR: activate/deactivate CDRs;
SMS-MT-CDR: active/deactivate CDRs.
The Charging Characteristics field allows the operator to apply different kind of charging methods in the CDRs. A subscriber may have Charging Characteristics assigned to his subscription. These characteristics can be supplied by the HLR to the SGSN as part of the subscription information, and, upon activation of a PDP context, the SGSN forwards the charging characteristics to the GGSN on the Gn / Gp reference point according to the rules specified in Annex A of TS 32.251 [11].
This information can be used by the GSNs to activate CDR generation and control the closure of the CDR or the traffic volume containers (see clause 5.1.2.2.23) and is included in CDRs transmitted to nodes handling the CDRs via the Ga reference point. It can also be used in nodes handling the CDRs (e.g., the CGF or the billing system) to influence the CDR processing priority and routing.
These functions are accomplished by specifying the charging characteristics as sets of charging profiles and the expected behaviour associated with each profile.
The interpretations of the profiles and their associated behaviours can be different for each PLMN operator and are not subject to standardisation. In the present document only the charging characteristic formats and selection modes are specified.
The functional requirements for the Charging Characteristics as well as the profile and behaviour bits are further defined in normative Annex A of TS 32.251 [11], including the definitions of the trigger profiles associated with each CDR type.
The format of charging characteristics field is depicted in Figure 4. Px (x =0..3) refers to the Charging Characteristics Profile index. Bits classified with a “B” may be used by the operator for non-standardised behaviour (see Annex A of TS 32.251 [11]).
Right, well hopefully next time someone goes looking for this info you’ll find it a bit more easily than I did!
SGs-AP which is used for CSFB & SMS doesn’t span network borders (you can’t roam with SGs-AP), and with SMSoIP out of the question, that gave us the option of MAP or Diameter, so we picked Diameter.
This introduces the S6c and SGd Diameter interfaces, in the diagrams below Orange is the Home Network (HPMN) and the Green is the Visited Network (VPMN).
The S6c interface is used between the SMSc and the HSS, in order to retrieve the routing information. This like the SRI-for-SM in MAP.
The SGd interface is used between the MME serving the UE and the SMSc, and is used for actual delivery of the MO/MT messages.
I haven’t shown the Diameter Routing Agents in these diagrams, but in reality there would be a DRA on the VPLMN and a DRA on the HPMN, and probably a DRA in the IPX between them too.
The Attach
The attach looks like a regular roaming attach, the MME in the Visited PMN sends an Update Location Request to the HSS, so the HSS knows the MME that is serving the subscriber.
S6a Update Location Request to indicate the MME serving the Subscriber
The Mobile Terminated SMS Flow
Now we introduce the S6c interface and the SGd interfaces.
When the Home SMSc has a message to send to the subscriber (Mobile Terminated SMS) it runs a the Send-Routing-Info-for-SM-Request (SRR) dialog to the HSS.
The Send-Routing-Info-for-SM-Answer (SRA) back from the HSS contains the info on the MME Diameter Host name and Diameter Realm serving the subscriber.
S6t – Send-Routing-Info-for-SM request to get the MME serving the subscriber
With this info, we can now craft a Diameter Request that will get sent to the MME serving the subscriber, containing the SMS PDU to send to the UE.
SGd MT-Forward-Short-Message to deliver Mobile Terminated SMS to the serving MME
We make sure it’s sent to the correct MME by setting the Destination-Host and Destination-Realm in the Diameter request.
Here’s how the request looks from the SMSc towards our DRA:
As you can see the Destination Realm and Destination-Host is set, as is the User-Name set to the IMSI of the UE we want to send the message to.
And down the bottom you can see the SMS-TPDU, the same as it’s been all the way back since GSM days.
The Mobile Originated SMS Flow
The Mobile Originated flow is even simpler, because we don’t need to look up where to route it to.
The MME receives the MO SMS from the UE, and shoves it into a Diameter message with Application ID set to SGd and Destination-Realm set to the HPMN Realm.
When the message reaches the DRA in the HPMN it forwards the request to an SMSc and then the Home SMSc has the message ready to roll.
To make more Money (This post, congratulations, you’re reading it!)
Because they have to (Regulatory compliance, insurance, taxes, etc) – That’s the next post
So let’s look at SA in this context.
5G-SA can drive new revenue streams
We (as an industry) suck at this.
Last year on the Telecoms.com podcast, Scott Bicheno made the point that if operators took all the money they’d gambled (and lost) on trying to play in the sports rights, involvement in media companies, building their own streaming apps, attempts at bundling other utilities, digital identity, etc, and just left the cash in the bank and just operated the network, they’d be better off.
Uber, Spotify, “OTTs”, etc, utilize MNOs to enable their services, but operators don’t see this extra revenue. While some operators may talk of “fair share” the truth is, these companies add value to our product (connectivity) which as an industry, we’ve failed to add ourselves.
If the Metaverse does turn out to be a cash cow, it is unlikely the telecommunications industry will be the ones milking it.
Claim: Customers are willing to pay more for 5G-SA
This myth seems to be fairly persistent, but with minimal data to support this claim.
While BSS vendors talk about “5G Monetization”, the truth is, people use their MNO to provide them connectivity. If the coverage is adequate, and the speed enough to do what they need to do, few would be willing to pay any additional cash each month to see higher numbers on a speedtest result (enabled by 5G-NSA) and even fewer would pay extra cash for, well, whatever those features only enabled by 5G-Standalone are?
With most consumers now also holding onto their mobile devices for longer periods of time, and with interest rates reining in consumer spending across the board, we are seeing the rise of a more cost conscious consumer than ever before. If we want to see higher ARPUs, we need to give the consumer a compelling reason to care and spend their cash, beyond a speed test result.
We talk a little about APIs lower down in the post.
Claim: Users want Ultra-Low Latency / High Reliability Comms that only 5G-SA delivers
Wanting to offer a product to the market, is not the same as the market wanting a product to consume.
Telecom operators want customers to want these services, but customer take up rates tell a different story. For a product like this to be viable, it must have a wide enough addressable market to justify the investment.
Reliability
The URLCC standards focus on preventing packet loss, but the world has moved on from needing zero packet loss.
The telecom industry has a habit of deciding what customers want without actually listening. When a customer talks about wanting “reliable” comms, they aren’t saying they want zero packet loss, but rather fewer dropouts or service flaps. For us to give the customer what they are actually asking for involves us expanding RAN footprint and adding transmission diversity, not 5G-SA.
The “protocols of the internet” (TCP/IP) have been around for more than 50 years now.
These protocols have always flowed over transport links with varied reliability and levels of packet loss.
Thanks to these error correction and retransmission techniques built into these protocols, a lost packet will not interrupt the stream. If your nuclear command and control network were carried over TCP/IP over the public internet (please don’t do this), a missing packet won’t lead to worldwide annihilation, but rather the sender will see the receiver never acknowledged the receipt of the packet at the other end, and resend it, end of.
If you walk into a hospital today, you’ll find patient monitoring devices, tracking the vital signs for patients and alerting hospital staff if a patient’s vital signs change. It is hard to think of more important services for reliability than this.
And yet they use WiFi, and have done for a long time, if a packet is lost on WiFi (as happens regularly) it’s just retransmitted and the end user never knows.
Autonomous cars are unlikely to ever rely on a 5G connection to operate, for the simple reason that coverage will never be 100%. If your car stops because you’re in a not-spot, you won’t be a happy customer. While plenty of cars have cellular modems in them, that are used to upload telemetry data back to the manufacturer, but not to drive the car.
One example of wireless controlled vehicles in the wild is autonomous haul trucks in mines. Historically, these have used WiFi for their comms. Mine sites are often a good fit for Private LTE, but there’s nothing inherent in the 5G Standalone standard that means it’s the only tool for the job here.
Slicing
Slicing is available in LTE (4G), with an architecture designed to allow access to others. It failed to gain traction, but is in networks today.
The RAN a piece of the latency puzzle here, but it is just one piece of the puzzle.
If we look at the flow a packet takes from the user’s device to the server they want to talk to we’ve got:
Time it takes the UE to craft the packet
Time it takes for the packet to be transmitted over the air to the base station
Time it takes for the packet to get through the RAN transmission network to the core
Time it takes the packet to traverse the packet core
Time it takes for the packet to get out to transit/peering
Time it takes to get the packet from the edge of the operators network to the edge of the network hosting the server
Time it takes the packet through the network the server is on
Time it takes the server to process the request
The “low latency” bit of the 5G puzzle only involves the two elements in bold.
If you’ve got to get from point A to point B along a series of roads, and the speed limit on two of the roads you traverse (short sections already) is increased. The overall travel time is not drastically reduced.
I’m lucky, I have access to a well kitted out lab which allows me to put all of these latency figures to the test and provide side by side metrics. If this is of interest to anyone, let me know. Otherwise in the meantime you’ll just have to accept some conjecture and opinion.
You could rebut this talking about Edge Compute, and having the datacenter at the base of the tower, but for a number of fairly well documented reasons, I think this is unlikely to attract widespread deployment in established carrier networks, and Intel’s recent yearly earning specifically called this out.
Claim: Customers want APIs and these needs 5G SA
Companies like Twilio have made it easy to interact with the carrier network via their APIs, but yet again, it’s these companies producing the additional value on a service operated by the MNOs.
My coffee machine does not have an API, and I’m OK with this because I don’t have a want or need to interact with it programatically.
By far, the most common APIs used by businesses involving telco markets are APIs to enable sending an SMS to a user.
These have been around for a long time, and the A2P market is pretty well established, and the good news is, operators already get a chunk of this pie, by charging for the SMS.
Imagine a company that makes medical booking software. They’re a tech company, so they want their stack to work anywhere in the world, and they want to be able to send reminder SMS to end users.
They could get an account manager with each of the telcos in each of the markets they work in, onboard and integrate the arcane complexities of each operators wholesale SMS system, or they could use Twilio or a similar service, which gives them global reach.
Often the cost of services like Twilio are cheaper than working directly with the carriers in each market, and even if it is marginally more expensive, the cost savings by not having to deal with dozens of carriers or integrate into dozens of systems, far outweighs this.
While it’s a great idea, in the context of 5G Standalone and APIs, it’s worth noting that none of the use cases in OpenGateway require 5G Standalone (Except possibly Edge discovery, but it is debatable).
Critically, from a developer experience perspective:
I can sign up to services like Twilio without a credit card, and start using the service right away, with examples in my programming language of choice, the developer user experience is fantastic.
Jump on the OpenGateway website today and see if you can even find a way to sign up to use the service?
Claim: Fixed Wireless works best with 5G-SA
Of all the touted use cases and applications for 5G, Fixed Wireless (FWA) has been the most successful.
The great thing about FWA on Cellular networks is you can use the same infrastructure you use for your mobile customers, and then sell excess capacity in the network to deliver Fixed Wireless Access services, better utilizing an asset (great!).
But again, this does not require Standalone 5G. If you deploy your FWA network using 5G SA, then you won’t be able to sweat that same asset for both mobile subscribers and FWA subscribers.
Today at least, very few handsets short of this generation of flagship phones, supports 5G SA. Even the phones sold as supporting 5G over the past few years, are almost all only supporting 5G-NSA, so if you rolled out your FWA network as Standalone, you can’t better utilize the asset by sharing with your existing LTE/5G-NSA customers.
Claim: The Killer App is coming for 5G and it needs 5G SA
This space is reserved for the killer app that requires 5G Standalone.
Whenever that comes?
Anyone?
I’m not paying to build a marina berth for my mega yacht, mostly because I don’t have one. Ditto this.
Could you explain to everyone on an investor call that you’re investing in something where the vessel of the payoff isn’t even known to exist? Telecom is “blue chip”, hardly speculative.
The Future for Revenue Growth?
Maybe there isn’t one.
I know it’s an unthinkable thought for a lot of operators, but let’s look at it rationally; in the developed world, everyone who wants a mobile service already has one.
This leaves operators with two options; gaining market share from their competitors and selling more/higher priced services to existing customers.
You don’t steal away customers from other operators by offering a higher priced product, and with reduced consumer spending people aren’t queuing up to spend more each month.
But there is a silver lining, if you can’t grow revenues, you can still shrink expenditure, which in the end still gets the same result at the end of the quarter – More cash.
Simplify your operations, focus on what you do really well (mobile services), the whole 80/20 rule, get better at self service, all that guff.
There’s no shortage of pain points for consumers telecom operators could address, to make the customer experience better, but few that include the word Slicing.
No one spends marketing dollars talking about the problems with a tech and vendors aren’t out there promoting sweating existing assets. But understanding your options as an operator is more important now than ever before.
Sidebar; This post got really long, so I’m splitting it into 3…
We’re often asked to help define a a 5G strategy for operators; while every case is different, there’s a lot of vendors pushing MNOs to move towards 5G standalone or 5G-SA.
I’m always a fan of playing “devil’s advocate“, and with so many articles and press releases singing the praises of standalone 5G/5G-SA, so as a counter in this post, I’ll be making the case against the narratives presented to operators by vendors that the “right” way to do 5G is to introduce 5G Standalone, that they should all be “upgrading” to Standalone 5G.
With Mobile World Congress around the corner, now seems like a good time to put forward the argument against introducing 5G Standalone, rebutting some common claims about 5G Standalone operators will be told. We’ll counterpoint these arguments and I’ll put forward the case for not jumping onto the 5G-SA bandwagon – just yet.
On a personal note, I do like 5G SA, it has some real advantages and some cool features, which are well documented, including on this blog. I’m not looking to beat up on any vendors, marketing hype or events, but just to provide the “other side” of the equation that operators should consider when making decisions and may not be aware of otherwise. It’s also all opinion of course (cited where possible), but if you’re going to build your network based on a blog post (even one as good as this) you should probably reconsider your life choices.
Some Arcane Detail: 5G Non-Standalone (NSA) vs Standalone (SA)
5G NSA (Non Standalone) uses LTE (4G) with an additional layer “bolted on” that uses 5G on the radio interface to provide “5G” speeds to users, while reusing the existing LTE (Evolved Packet Core) core and VoLTE for voice / SMS.
From an operator perspective there is almost no change required in the network to support NSA 5G, other than in the RAN, and almost all the 5G networks in commercial use today use 5G NSA.
5G NSA is great, it gives the user 5G speeds for users with phones that support it, with no change to the rest of the network needed.
Standalone 5G on the other hand requires an a completely new core network with all the trimmings.
While it is possible to handover / interwork with LTE/4G (Inter-RAT Handovers), this is like 3G/4G interworking, where each has a different core network. Introducing 5G standalone touches every element of the network, you need new nodes supporting the new standards for charging, policy, user plane, IMS, etc.
Scope
There’s an old adage that businesses spend money for one of three reasons:
To Save Money (Which we’ll cover in this post)
To make more Money (Covered next – Will link when published)
Because they have to (Regulatory compliance, insurance, taxes, etc)
Let’s look at 5G Standalone in each of these contexts:
5G Cost Savings – Counterpoint: The cost-benefit doesn’t stack up
As an operator with an existing deployed 4G LTE network, deploying a new 5G standalone network will not save you money.
From an capital perspective this is pretty obvious, you’re going to need to invest in a new RAN and a new core to support this, but what about from an opex perspective?
Claim: 5G RAN is more efficient than 4G (LTE) RAN
Spectrum is both finite and expensive, so MNOs must find the most efficient way to use that spectrum, to squeeze the most possible value out of it.
In rough numbers, we can say we get 5x the spectral efficiency by moving from 3G to 4G. This means we can carry 5.2x more with the same spectrum on 4G than we can on 3G – A very compelling reason to upgrade.
The like-for-like spectral efficiency of 5G is not significantly greater than that of LTE.
In numbers the same 5Mhz of spectrum we refarmed from UMTS (3G) to 4G (LTE) provided a 5x gain in efficiency to deliver 75Mbps on LTE. The same configuration refarmed to 5G-NR would provide 80Mbps.
Refarming spectrum from 4G (LTE) to 5G (NR) only provides a 6% increase in spectral efficiency.
While 6% is not nothing, if refarmed to a 5G standalone network, the spectrum can no longer be used by LTE only devices (Unless Dynamic Spectrum Sharing is used which in itself leads to efficiency losses), which in itself reduces the efficiency and would add additional load to other layers.
The crazy speeds demonstrated by 5G are not due to meaningful increases in efficiency, but rather the ability to use more spectrum, spectrum that operators need to purchase at auction, purchase equipment to utilize and pay to run.
Claim: 5G Standalone Core is Cheaper to operate as it is “Cloud Native”
It has been widely claimed that the shift for the 5G Core Architecture to being “Cloud Native” can provide cost savings.
Operators should regard this in a skeptical manner; after all, we’ve been here before.
Did moving from big-iron to VNFs provide the promised cost savings to operators?
For many operators the shift from hardware to software added additional complexity to the network and increased the headcount to support this.
What were once big-iron appliances dedicated to one job, that sat in the corner and chugged away, are now virtual machines (VNFs). Many operators have naturally found themselves needing a larger team to manage the virtual environment, compared to the size of the team they needed to just to plug power and data into a big box in an exchange before everything was virtualized.
Introducing a “Cloud Native” Kubernetes layer on top of the VNF / virtualization layer, on top of the compute layer, leaves us with a whole lot of layers. All of which require resources to be maintain, troubleshoot and kept running; each layer having associated costs for staffing, licensing and support.
Many mid size enterprises rushed into “the cloud” for the promised cost savings only to sheepishly admit it cost more than the expected.
Almost none of the operators are talking about running these workloads in the public cloud, but rather “Private Clouds” built on-premises, using “Cloud Native” best practices.
One of the central arguments about cloud revolves around “elastic scaling” where the network can automatically scale to match demand; think extra instances spun up a times of peak demand and shut down when the demand drops.
I explain elastic scaling to clients as having to move people from one place to another. Most of the time, I’m just moving myself, a push bike is fine, or I’ve got a 4 seater car, but occasionally I’ll need to move 25 people and for that I’d need a bus.
If I provide the transportation myself, I need to own a bike, a car and a bus.
But if use the cloud I can start with the push bike, and as I need to move more people, the “cloud” will provide me the vehicle I need to move the people I need to move at that moment, and I’ll just pay for the time I need the bus, and when I’m done needing the bus, I drop back to the (cheaper) push bike when I’m not moving lots of people.
While telecom operators are going to provide the servers to run this in “On-prem-cloud”, they need to dimension for the maximum possible load. This means they need to own a bike/car/bus, even if they’re not using it most of the time, and there’s really no cost savings to having a bus but not using it when you’re not paying by the hour to hire it.
Infrastructure aside, introducing a Standalone 5G Core adds another core network to maintain. Alongside the Circuit Switched Core (MSC/GGSN/SGSN) serving 2G/3G subscribers, Evolved Packet Core serving 4G (LTE) and 5G-NSA subscribers, adding a 5G Standalone Core to for the 5G-SA subscribers served by the 5G SA cells, is going to be more work (and therefore cost).
While the majority of operators have yet to turn off their 2G/3G core networks, introducing another core network to run in parallel is unlikely to lead to any cost savings.
Claim: Upgrading now can save money in the Future / Future Proofing
Life cycles of telecommunications are two fold, one is the equipment/platform life cycle (like the RAN components or Core network software being used to deliver the service) the other is the technology life cycle (the generation of technology being used).
The technology lifecycles in telecommunications are vastly longer than that for regular tech.
GSM (2G) was introduced into the UK in 1991, and will be phased out starting in 2033, a 42 year long technology life cycle.
No vendor today could reasonably expect the 5G hardware you deploy in 2024 to still be in production in 2066 – The platform/equipment life cycle is a lot shorter than the technology life cycle.
Operators will to continue relying on LTE (4G) well into the late 2030s.
I’d wager that there is not a single piece of equipment in the Vodafone UK GSM network today, that was there in 1991. I’d go even further to say that any piece of equipment in the network today, didn’t even replace the 1991 equipment, but was probably 3 or 4 generations removed from the network built in 1991.
For most operators, RAN replacements happen between 4 to 7 years, often with targeted augmentation / expansion as needed in the form of adding extra layers / sectors between these times.
The question operators should be asking is therefore not what will I need to get me through to 2066, but rather what will I need to get to 2030?
The majority of operators outside the US today still operate a 2G or 3G network, generally with minimal bandwidth to support legacy handsets and devices, while the 4G (LTE) network does most of the heavy lifting for carrying user traffic. This is often with the aid of an additional 5G-NSA (Non-Standalone) layer to provide additional capacity.
Is there a cost saving angle to adding support for 5G-Standalone in addition to 2G/3G/4G (LTE) and 5G (Non-Standalone) into your RAN?
A logical stance would be that removing layers / technologies (such as 2G/3G sunsetting) would lead to cost savings, and adding a 5G Standalone layer would increase cost.
All of the RAN solutions on the market today from the major vendors include support for both Standalone 5G and Non Standalone, but the feature licensing for a non-standalone 5G is generally cheaper than that for Standalone 5G.
The question operators should be asking is on what timescale do I need Standalone 5G?
If you’ve rolled out 5G-NSA today, then when are you looking to sunset your LTE network? If the answer is “I hope to have long since retired by that time”, then you’ve just answered that question and you don’t need to licence / deploy 5G-SA in this hardware refresh cycle.
Other Cost Factors
Roaming: The majority of roaming traffic today relies on 2G/3G for voice. VoLTE roaming is (finally) starting to establish a foothold, but we are a long way from ubiquitous global roaming for LTE and VoLTE, and even further away for 5G-SA roaming. Focusing on 5G roaming will enable your network for roaming use by a miniscule number of operators, compared to LTE/VoLTE roaming which covers the majority of the operators in the developed world who can utilize your service.
I decided to split this into 3 posts, next I’ll post the “5G can make us more money” post and finally a “5G because we have to” post. I’ll post that on LinkedIn / Twitter / Mailing list, so stick around, and feel free to trash me in the comments.
In our last post we looked at Actions and ActionPlans, and one of the really funky things we can do is setting ActionPlans to trigger on a time schedule or setting ActionTriggers to trigger on an event.
We’re going to build on the examples we had on the last post, so we’ll assume your code is up to the point where we’ve added a Signup Bonus to an account, using an ActionPlan we assigned when creating the account.
In this post, we’re going to create an action that charges $6, called “Action_Monthly_Charge“, and tie it to an ActionPlan called “ActionPlan_Monthly_Charge“, but to demo how this works rather than charging this Monthly, we’re going to charge it every minute.
Then with our balances ticking down, we’ll set up an ActionTrigger to trigger when the balance drops below $95, and alert us.
Defining the Monthly Charge Action
The Action for the Monthly charge will look much like the other actions we’ve defined, except the Identifier is *debitso we know we’re deducting from the balance, and we’ll log to the CDRs table too:
Next we’ll need to wrap this up into an ActionPlan, this is where some of the magic happens. Inside the action plan we can set a once off time, or a recurring time, kinda like Cron.
We’re setting the time to *every_minute so things will happen quickly while we watch, this action will get triggered every 60 seconds. In real life of course, for a Monthly charge, we’d want to trigger this Action monthly, so we’d set this value to *monthly. If we wanted this to charge on the 2nd of the month we’d set the MonthDays to “2”, etc, etc.
If you think the accounts will start getting debited every 60 seconds after applying this, you’d be wrong, we need to associate this ActionPlan with an Account first, this is how we control which accounts get which ActionPlans tied to them, to do this we’ll use the SetAccout API again we’ve been using to create accounts:
Well, for starters the ActionPlan named “ActionPlan_Signup_Bonus” is going to be triggered, as in the ActionPlan it’s Timing is set to *asap, so CGrateS will apply the corresponding Action (“Action_Add_Signup_Bonus“) right away, which will credit the account $99.
But a minute after that, we’ll trigger the ActionPlan named “ActionPlan_Monthly_Charge”, as the timing for this is set to *every_minute, when the Action “Action_Monthly_Charge” is triggered, it’s going to be deducting $6 from the balance.
We can check this by using the GetAccount API:
# Get Account Info
pprint.pprint(CGRateS_Obj.SendData({'method': 'ApierV2.GetAccount', 'params': [
{"Tenant": "cgrates.org", "Account": str(Account)}]}))
You should see a balance of $99 to start with, and then after 60 seconds, it should be down to $93, and so on.
Triggering Actions based on Balances with ActionTriggers
Okay, so we’ve set up recurring charges, now let’s get notified if the balance drops below $95, we’ll start, like we have before, with defining an Action, this will log to the CDRs table, HTTP post and write to syslog:
Now we’ll define an ActionTrigger to check if the balance is below $95 and trigger our newly created Action (“Action_HTTP_Notify_95“) when that condition is met:
We’ve defined the ThresholdType of *min_balance, but we could equally set this to ThresholdType to *max_balance, *balance_expired or trigger when a certain Counter has been triggered enough times.
Adding an ActionTrigger to an Account
Again, like the ActionPlan we created before, before the ActionTrigger we just created will be used, we need to associate it with an Account, for this we’ll use the AddAccountActionTriggers API, specify the Account and the ActionTriggerID for the ActionTrigger we just created.
If we run this all together, creating the account with the “ActionPlan_Signup_Bonus” will give the account a $99 Balance. But after 60 seconds, “ActionPlan_Monthly_Charge” will kick in, and every 60 seconds after that, at which point the balance will get to below $95 when CGrateS will trigger the ActionTrigger “ActionTrigger_95_Remaining” and get the HTTP POST to the HTTP endpoint and log entry:
We can check on this using the ApierV2.GetAccount method, where we’ll see the ActionTrigger we just defined.
Checking out the LastExecutionTime we can see if the ActionTrigger been triggered or not.
So using this technique, we can notify a customer when they’ve used a certain amount of their balance, but we can lock out Accounts who have spent more than their allocated spend limit by setting an Action that suspends the Account once it reaches a certain level. We notify customers when balance expires, or if a certain number of counters has been triggered.
In the cellular world, subscribers are charged for data from the IP, transport and applications layers; this means you pay for the IP header, you pay for the TCP/UDP header, and you pay for the contents (the cat videos it contains).
This also means if an operator moves mobile subscribers from IPv4 to IPv6, there’s an extra 20 bytes the customer is charged for for every packet sent / received, which the customer is charged for – This is because the IPv6 header is longer than the IPv4 header.
In most cases, mobile subs don’t get a choice as to if their connection is IPv4 or IPv6, but on a like for like basis, we can say that if a customer moves is on IPv6 every packet sent/received will have an extra 20 bytes of data consumed compared to IPv4.
This means subscribers use more data on IPv6, and this means they get charged for more data on IPv6.
For IoT applications, light users and PAYG users, this extra 20 bytes per packet could add up to something significant – But how much?
We can quantify this, but we’d need to know the number of packets sent on average, and the quantity of the data transferred, because the number of packets is the multiplier here.
So for starters I’ve left a phone on the desk, it’s registered to the network but just sitting in Idle mode – This is an engineering phone from an OEM, it’s just used for testing so doesn’t have anything loaded onto it in terms of apps, it’s not signed into any applications, or checking in the background, so I thought I’d try something more realistic.
So to get a clearer picture, I chucked a SIM in my regular everyday phone I use personally, registered it to the cellular lab I have here. For the next hour I sniffed the GTP traffic for the phone while it was sitting on my desk, not touching the phone, and here’s what I’ve got:
Overall the PCAP includes 6,417,732 bytes of data, but this includes the transport and GTP headers, meaning we can drop everything above it in our traffic calculations.
For this I’ve got 14 bytes of ethernet, 20 bytes IP, 8 bytes UDP and 5 bytes for TZSP (this is to copy the traffic from the eNB to my local machine), then we’ve got the transport from the eNB to the SGW, 14 bytes of ethernet again, 20 bytes of IP , 8 bytes of UDP and 8 bytes of GTP then the payload itself. Phew. All this means we can drop 97 bytes off every packet.
We have 16,889 packets, 6,417,732 bytes in total, minus 97 bytes from each gives us 1,638,233 of headers to drop (~1.6MB) giving us a total of 4.556 MB traffic to/from the phone itself.
This means my Android phone consumes 4.5 MB of cellular data in an hour while sitting on the desk, with 16,889 packets in/out.
Okay, now we’re getting somewhere!
So now we can answer the question, if each of these 16k packets was IPv6, rather than IPv4, we’d be adding another 20 bytes to each of them, 20 bytes x 16,889 packets gives 337,780 bytes (~0.3MB) to add to the total.
If this traffic was transferred via IPv6, rather than IPv4, we’d be looking at adding 20 bytes to each of the 16,889 packets, which would equate to 0.3MB extra, or about 7% overhead compared to IPv4.
But before you go on about what an outrage this IPv6 transport is, being charged for those extra bytes, that’s only one part of the picture.
There’s a reason operators are finally embracing IPv6, and it’s not to put an extra 7% of traffic on the network (I think if you asked most capacity planners, they’d say they want data savings, not growth).
IPv6 is, for lack of a better term, less rubbish than IPv4.
There’s a lot of drivers for IPv6, and some of these will reduce data consumption. IPv6 is actually your stuff talking directly to the remote stuff, this means that we don’t need to rely on NAT, so no need to do NAT keepalives, and opening new sessions, which is going to save you data. If you’re running apps that need to keep a connection to somewhere alive, these data savings could negate your IPv6 overhead costs.
Will these potential data savings when using IPv6 outweigh the costs?
That’s going to depend on your use case.
If you’ve extremely bandwidth / data constrained, for example, you have an IoT device on an NTN / satellite connection, that was having to Push data every X hours via IPv4 because you couldn’t pull data from it as it had no public IP, then moving it to IPv6 so you can pull the data on the public IP, on demand, will save you data. That’s a win with IPv6.
If you’re a mobile user, watching YouTube, getting push notifications and using your phone like a normal human, probably not, but if you’re using data like a normal user, you’ve probably got a sizable data allowance that you don’t end up fully consuming, and the extra 20 bytes per packet will be nothing in comparison to the data used to watch a 2k video on your small phone screen.
Even before 5G was released, the arms race to claim the “fastest” speeds on LTE, NSA and SA networks has continued, with pretty much every operator claiming a “first” or “fastest”.
I myself have the fastest 5G network available* but I thought I’d look at how big the values are we can put in for speed, these are the Maximum Bitrate Values (like AMBR) we can set on an APN/DNN, or on a Charging Rule.
*Measurement is of the fastest 5G network in an eastward facing office, operated by a person named Nick, in a town in Australia. Other networks operated by people other than those named Nick in eastward facing office outside of Australia were not compared.
The answer for Release 8 LTE is 4294967294 bytes per second, aka 4295 Mbps 4.295 Gbps.
Not bad, but why this number?
The Max-Requested-Bandwidth-DL AVP tells the PGW the max throughput allowed in bits per second. It’s a Unsigned32 so max value is 4294967294, hence the value.
But come release 15 some bright spark thought we may in the not to distant future break this barrier, so how do we go above this?
The answer was to bolt on another AVP – the “Extended-Max-Requested-BW-DL” AVP ( 554 ) was introduced, you might think that means the max speed now becomes 2x 4.295 Gbps but that’s not quite right – The units was shifted.
This AVP isn’t measuring bits per second it’s measuring kilobits per second.
So the standard Max-Requested-Bandwidth-DL AVP gives us 4.3 Gbps, while the Extended-Max-Requested-Bandwidth gives us a 4,295 Gbps.
We add the Extended-Max-Requested-Bandwidth AVP (4295 Gbps) onto the Max-Requested Bandwidth AVP (4.3 Gbps) giving us a total of 4,4299.3 Gbps.
The Binding Support Function is used in 4G and 5G networks to allow applications to authenticate against the network, it’s what we use to authenticate for XCAP and for an Entitlement Server.
Rather irritatingly, there are two BSF addresses in use:
If the ISIM is used for bootstrapping the FQDN to use is:
bsf.ims.mncXXX.mccYYY.pub.3gppnetwork.org
But if the USIM is used for bootstrapping the FQDN is
bsf.mncXXX.mccYYY.pub.3gppnetwork.org
You can override this by setting the 6FDA EF_GBANL (GBA NAF List) on the USIM or equivalent on the ISIM, however not all devices honour this from my testing.
One of the hyped benefits of a 5G Core Networks is that 5GC can be used for wired networks (think DSL or GPON) – In marketing terms this is called “Wireless Wireline Convergence” (5G WWC) meaning DSL operators, cable operators and fibre network operators can all get in on this sweet 5GC action and use this sexy 5G Core Network tech.
This is something that’s in the standards, and that the big kit vendors are pushing heavily in their marketing materials. But will it take off? And should operators of wireline networks (fixed networks) be looking to embrace 5GC?
Comparing 5GC with current wireline network technologies isn’t comparing apples to apples, it’s apples to oranges, and they’re different fruits.
At its heart, the 3GPP Core Networks (including 5G Core) address one particular use cases of the cellular industry: Subscriber mobility – Allowing a customer to move around the network, being served by different kit (gNodeBs) while keeping the same IP Address.
The most important function of 5GC is subscriber mobility.
This is achieved through the use of encapsulating all the subscriber’s IP data into a GTP (A protocol that’s been around since 2G first added data).
Do I need a 5GC for my Fixed Network?
Wireline networks are fixed. Subscribers don’t constantly move around the network. A GPON customer doesn’t need to move their OLT every 30 minutes to a new location.
Encapsulating a fixed subscriber’s traffic in GTP adds significant processing overhead, for almost no gain – The needs of a wireline network operator, are vastly different to the needs of a cellular core.
Today, you can take a /24 IPv4 block, route it to a DSLAM, OLT or CMTS, and give an IP to 254 customers – No cellular core needed, just a router and your access device and you’re done, and this has been possible for decades. Because there’s no mobility the GTP encapsulation that is the bedrock for cellular, is not needed.
Rather than routing directly to Access Network kit, most fixed operators deploy BRAS systems used for fixed access. Like the cellular packet core, BRAS has been around for a very long time, with a massive install base and a sea of engineering experience in house, it meets the needs of the wireline industry who define its functions and roles along with kit vendors of wireline kit; the fixed industry working groups defined the BRAS in the same way the 3GPP and cellular industry working groups defined 5G Core.
I don’t forsee that we’ll see large scale replacement of BRAS by 5GC, for the same reason a wireless operator won’t replace their mobile core with a BRAS and PPPoE – They’re designed to meet different needs.
All the other features that have been added to the 3GPP Core Network functionality, like limiting speed, guaranteed throughput bearers, 5QI / QCI values, etc, are addons – nice-to-haves. All of these capabilities could be implemented in wireline networks today – if the business case and customer demand was there.
But what about slicing?
With dropping ARPUs across the board, additional services relating to QoS (“Network Slicing”) are being held up as the saving grace of revenues for cellular operators and 5G as a whole, however this has yet to be realized and early indications suggest this is not going to be anywhere near as lucrative as previously hoped.
What about cost savings?
In terms of cost-per-bit of throughput, the existing install base wireline operators have of heavy-metal kit capable of terabit switching and routing has been around for some time in fixed world, and is what most 5G Cores will connect to as their upstream anyway, so there won’t be any significant savings on equipment, power consumption or footprint to be gained.
Fixed networks transport the majority of the world’s data today – Wireline access still accounts for the majority of traffic volumes, so wireline kit handles a higher magnitude of throughput than it’s Packet Core / 5GC cousins already.
Cutting down the number of parts in the network is good though right?
If you’re operating both a Packet Core for Cellular, and a fixed network today, then you might think if you moved from the traditional BRAS architecture fore the wired network to 5GC, you could drop all those pesky routers and switches clogging up your CO, Exchanges and Data Centers.
The problem is that you still need all of those after the 5GC to be able to get the traffic anywhere users want to go. So the 5GC will still need all of that kit, all your border routers and peering routers will remain unchanged, as well as domestic transmission, MPLS and transport.
The parts required for operating fixed networks is actually pretty darn small in comparison to that of 5GC.
TL;DR?
While cellular vendors would love to sell their 5GC platform into fixed operators, the premise that they are willing to replace existing BRAS architectures with 5GC, is as unlikely in my view as 5GC being replaced by BRAS.
Misunderstood, under appreciated and more capable than people give it credit for, is our PCRF.
But what does it do?
Most folks describe the PCRF in hand wavy-terms – “it does policy and charging” is the answer you’ll get, but that doesn’t really tell you anything.
So let’s answer it in a way that hopefully makes some practical sense, starting with the acronym “PCRF” itself, it stands for Policy and Charging Rules Function, which is kind of two functions, one for policy and one for rules, so let’s take a look at both.
Policy
In cellular world, as in law, policy is the rules.
For us some examples of policy could be a “fair use policy” to limit customer usage to acceptable levels, but it can also be promotional packages, services like “free Spotify” packages, “Voice call priority” or “unmetered access to Nick’s Blog and maximum priority” packages, can be offered to customers.
All of these are examples of policy, and to make them work we need to target which subscribers and traffic we want to apply the policy to, and then apply the policy.
Charging Rules
Charging Rules are where the policy actually gets applied and the magic happens.
It’s where we take our policy and turn it into actionable stuff for the cellular world.
Let’s take an example of “unmetered access to Nick’s Blog and maximum priority” as something we want to offer in all our cellular plans, to provide access that doesn’t come out of your regular usage, as well as provide QCI 5 (Highest non dedicated QoS) to this traffic.
To achieve this we need to do 3 things:
Profile the traffic going to this website (so we capture this traffic and not regular other internet traffic)
Charge it differently – So it’s not coming from the subscriber’s regular balance
Up the QoS (QCI) on this traffic to ensure it’s high priority compared to the other traffic on the network
So how do we do that?
Profiling Traffic
So the first step we need to take in providing free access to this website is to filter out traffic to this website, from the traffic not going to this website.
Let’s imagine that this website is hosted on a single machine with the IP 1.2.3.4, and it serves traffic on TCP port 443. This is where IPFilterRules (aka TFTs or “Traffic Flow Templates”) and the Flow-Description AVP come into play. We’ve covered this in the past here, but let’s recap:
IPFilterRules are defined in the Diameter Base Protocol (IETF RFC 6733), where we can learn the basics of encoding them,
They take the format:
action dir proto from src to dst
The action is fairly simple, for all our Dedicated Bearer needs, and the Flow-Description AVP, the action is going to be permit. We’re not blocking here.
The direction (dir) in our case is either in or out, from the perspective of the UE.
Next up is the protocol number (proto), as defined by IANA, but chances are you’ll be using 17 (UDP) or 6 (TCP).
The from value is followed by an IP address with an optional subnet mask in CIDR format, for example from 10.45.0.0/16would match everything in the 10.45.0.0/16 network.
Following from you can also specify the port you want the rule to apply to, or, a range of ports.
Like the from, the tois encoded in the same way, with either a single IP, or a subnet, and optional ports specified.
And that’s it!
So let’s create a rule that matches all traffic to our website hosted on 1.2.3.4 TCP port 443,
permit out 6 from 1.2.3.4 443 to any 1-65535
permit out 6 from any 1-65535 to 1.2.3.4 443
All this info gets put into the Flow-Information AVPs:
With the above, any traffic going to/from 1.23.4 on port 443, will match this rule (unless there’s another rule with a higher precedence value).
Charging Actions
So with our traffic profiled, the next question is what actions are we going to take, well there’s two, we’re going to provide unmetered access to the profiled traffic, and we’re going to use QCI 4 for the traffic (because you’ll need a guaranteed bit rate bearer to access!).
Charging-Group for Profiled Traffic
To allow for Zero Rating for traffic matching this rule, we’ll need to use a different Rating Group.
Let’s imagine our default rating group for data is 10000, then any normal traffic going to the OCS will use rating group 10000, and the OCS will apply the specific rates and policies based on that.
Rating Groups are defined in the OCS, and dictate what rates get applied to what Rating Groups.
For us, our default rating group will be charged at the normal rates, but we can define a rating group value of 4000, and set the OCS to provide unlimited traffic to any Credit-Control-Requests that come in with Rating Group 4000.
This is how operators provide services like “Unlimited Facebook” for example, a Charging Rule matches the traffic to Facebook based on TFTs, and then the Rating Group is set differently to the default rating group, and the OCS just allows all traffic on that rating group, regardless of how much is consumed.
Inside our Charging-Rule-Definition, we populate the Rating-Group AVP to define what Rating Group we’re going to use.
Setting QoS for Profiled Traffic
The QoS Description AVP defines which QoS parameters (QCI / ARP / Guaranteed & Maximum Bandwidth) should be applied to the traffic that matches the rules we just defined.
As mentioned at the start, we’ll use QCI 4 for this traffic, and allocate MBR/GBR values for this traffic.
Putting it Together – The Charging Rule
So with our TFTs defined to match the traffic, our Rating Group to charge the traffic and our QoS to apply to the traffic, we’re ready to put the whole thing together.
So here it is, our “Free NVN” rule:
I’ve attached a PCAP of the flow to this post.
In our next post we’ll talk about how the PGW handles the installation of this rule.
Next we’ll need to define our rt_pyform config, this is a super simple 3 line config file that specifies the path of what we’re doing:
DirectoryPath = "." # Directory to search
ModuleName = "script" # Name of python file. Note there is no .py extension
FunctionName = "transform" # Python function to call
The DirectoryPath directive specifies where we should search for the Python code, and ModuleName is the name of the Python script, lastly we have FunctionName which is the name of the Python function that does the rewriting.
Now let’s write our Python function for the transformation.
The Python function much have the correct number of parameters, must return a string, and must use the name specified in the config.
The following is an example of a function that prints out all the values it receives:
Note the order of the arguments and that return is of the same type as the AVP value (string).
We can expand upon this and add conditionals, let’s take a look at some more complex examples:
def transform(appId, flags, cmdCode, HBH_ID, E2E_ID, AVP_Code, vendorID, value):
print('[PYTHON]')
print(f'|-> appId: {appId}')
print(f'|-> flags: {hex(flags)}')
print(f'|-> cmdCode: {cmdCode}')
print(f'|-> HBH_ID: {hex(HBH_ID)}')
print(f'|-> E2E_ID: {hex(E2E_ID)}')
print(f'|-> AVP_Code: {AVP_Code}')
print(f'|-> vendorID: {vendorID}')
print(f'|-> value: {value}')
#IMSI Translation - if App ID = 16777251 and the AVP being evaluated is the Username
if (int(appId) == 16777251) and int(AVP_Code) == 1:
print("This is IMSI '" + str(value) + "' - Evaluating transformation")
print("Original value: " + str(value))
value = str(value[::-1]).zfill(15)
The above look at if the App ID is S6a, and the AVP being checked is AVP Code 1 (Username / IMSI ) and if so, reverses the username, so IMSI 1234567 becomes 7654321, the zfill is just to pad with leading 0s if required.
Now let’s do another one for a Realm Rewrite:
def transform(appId, flags, cmdCode, HBH_ID, E2E_ID, AVP_Code, vendorID, value):
#Print Debug Info
print('[PYTHON]')
print(f'|-> appId: {appId}')
print(f'|-> flags: {hex(flags)}')
print(f'|-> cmdCode: {cmdCode}')
print(f'|-> HBH_ID: {hex(HBH_ID)}')
print(f'|-> E2E_ID: {hex(E2E_ID)}')
print(f'|-> AVP_Code: {AVP_Code}')
print(f'|-> vendorID: {vendorID}')
print(f'|-> value: {value}')
#Realm Translation
if int(AVP_Code) == 283:
print("This is Destination Realm '" + str(value) + "' - Evaluating transformation")
if value == "epc.mnc001.mcc001.3gppnetwork.org":
new_realm = "epc.mnc999.mcc999.3gppnetwork.org"
print("translating from " + str(value) + " to " + str(new_realm))
value = new_realm
else:
#If the Realm doesn't match the above conditions, then don't change anything
print("No modification made to Realm as conditions not met")
print("Updated Value: " + str(value))
In the above block if the Realm is set to epc.mnc001.mcc001.3gppnetwork.org it is rewritten to epc.mnc999.mcc999.3gppnetwork.org, hopefully you can get a handle on the sorts of transformations we can do with this – We can translate any string type AVPs, which allows for hostname, realm, IMSI, Sh-User-Data, Location-Info, etc, etc, to be rewritten.
NB-IoT introduces support for NIDD – Non-IP Data Delivery (NIDD) which is one of the cool features of NB-IoT that’s gaining more widespread adoption.
Let’s take a deep dive into NIDD.
The case against IP for IoT
In the over 40 years since IP was standardized, we’ve shoehorned many things onto IP, but IP was never designed or optimized for low power, low throughput applications.
For the battery life of an IoT device to be measured in years, it has to be very selective about what power hungry operations it does. Transmitting data over the air is one of the most power-intensive operations an IoT device can perform, so we need to do everything we can to limit how much data is sent, and how frequently.
Use Case – NB-IoT Tap
Let’s imagine we’re launching an IoT tap that transmits information about water used, as part of our revolutionary new “Water as a Service” model (WaaS) which removes the capex for residents building their own water treatment plant in their homes, and instead allows dynamic scaling of waterloads as they move to our new opex model.
If I turn on the tap and use 12L of water, when I turn off the tap, our IoT tap encodes the usage onto a single byte and sends the usage information to our rain-cloud service provider.
So we’re not constantly changing the batteries in our taps, we need to send this one byte of data as efficiently as possible, so as to maximize the battery life.
If we were to transport our data on TCP, well we’d need a 3 way handshake and several messages just to transmit the data we want to send.
Let’s see how our one byte of data would look if we transported it on TCP.
That sliver of blue in the diagram is our usage component, the rest is overhead used to get it there. Seems wasteful huh?
Sure, TCP isn’t great for this you say, you should use UDP! But even if we moved away from TCP to UDP, we’ve still got the IPv4 header and the UDP header wasting 28 bytes.
For efficiency’s sake (To keep our batteries lasting as long as possible) we want to send as few messages as possible, and where we do have to send messages, keep them very short, so IP is not a great fit here.
Enter NIDD – Non-IP Data Delivery.
Through NIDD we can just send the single hex byte, only be charged for the single hex byte, and only stay transmitting long enough to send this single byte of hex (Plus the NBIoT overheads / headers).
Compared to UDP transport, NIDD provides us a reduction of 28 bytes of overhead for each message, or a 96% reduction in message size, which translates to real power savings for our IoT device.
In summary – the more sending your device has to do, the more battery it consumes. So in a scenario where you’re trying to maximize power efficiency to keep your batter powered device running as long as possible, needing to transmit 28 bytes of wasted data to transport 1 byte of usable data, is a real waste.
Delivering the Payload
NIDD traffic is transported as raw hex data end to end, this means for our 1 byte of water usage data, the device would just send the hex value to be transferred and it’d pop out the other end.
To support this we introduce a new network element called the SCEF –Service Capability Exposure Function.
From a developer’s perspective, the SCEF is the gateway to our IoT devices. Through the RESTful API on the SCEF (T8 API), we can send and receive raw hex data to any of our IoT devices.
When one of our Water-as-a-Service Taps sends usage data as a hex byte, it’s the software talking on the T8 API to the SCEF that receives this data.
Data of course needs to be addressed, so we know where it’s coming from / going to, and T8 handles this, as well as message reliability, etc, etc.
This is a telco blog, so we should probably cover the MME connection, the MME talks via Diameter to the SCEF. In our next post we’ll go into these signaling flows in more detail.
If you’re wondering what the status of Open Source SCEF implementations are, then you may have already guessed I’m working on one!
Hopefully by now you’ve got a bit of an idea of how NIDD works in NB-IoT, and in our next posts we’ll dig deeper into the flows and look at some PCAPs together.
