4G/LTE - Private Network

 

 

 

Private Celluar Network

 

As far as I remember, the first time I saw a Private LTE Network solution was at CTIA show in Sep 2015. I don't recall the name of the company for now but there was a booth presenting a small femto cell sized box which contains an LTE dNB and EPC (core network) in the single box. The person in the booth said the product is good for building my own LTE network with a single box and I can get it connected to the internal with a ethernet port.

At the time I was interested in the box only for technical reason. I wish I could get one of those boxes because I could play with various things on both eNB and core network of LTE and try many experiments. But I couldn't think of strong business drive because I thought there would be many obstacles to build a private network with a cellular technology like LTE. However, recently (as of late 2017) I started seeing more and more solutions for private LTE network and also see some movement to remove some of the obstacles that I thought of.

How can I get the spectrum to deploy the private network ?

The first question that struck me when I first saw the eNB+EPC one box was 'in which spectrum (frequency) I can use this box'.  Of course the first answer that I could think of was to use ISM band in which we normally use Bluetooth and WiFi. The frequency may work if I configure this box within a very short radius (like only within my house) and replacing the WiFi AP(Access Point) in my house. However, it is too costly to use the solution only to cover a couple of devices in my house and it would easily interfere the WiFi operation in the neigbours. It would sound more reasonable if we use the solution to cover a wider area like a whole campus and office building etc. However, even in this case using the ISM band would be a little bit risky to interfere WiFi in the neighbour.

Luckily recently FCC freed up pretty wide spectrum mainly for this kind of private network purpose. The band is called CBRS. It is likely to see this kind of free spectrum for other regions as well.

Three routes to spectrum exist, and two of them appear above. The first is unlicensed spectrum, which anyone may use and nobody may protect. The second is shared spectrum such as CBRS, where a coordinator grants the right to transmit. The third is licensed spectrum, which has to be obtained from whoever holds the licence.

The objection raised above to the ISM band is worth separating into its two halves. Using it is legal, and that is not the problem. The problem is that nothing stops a neighbour from using the same channel, so a private network built there has no way to protect the service level it was meant to guarantee.

  • Unlicensed spectrum gives no protection : the difficulty is coexistence rather than permission, and a mission critical deployment is exactly the case that cannot tolerate it.
  • Shared spectrum was the answer for LTE : CBRS provides a coordinated right to transmit, which is what the ISM band lacks.
  • Licensed spectrum stays with its holder : it can be used for a private network only with the agreement of the operator that holds it.
  • The regulator moved before 3GPP did : the spectrum question was answered by the FCC, and the protocol question waited for Release 16.

Who is going to manufacture the device (UE) ?

Another question that popped up in my mind when I first saw this solution was "OK, it is good to have this kind of small sized / private network solution. However, who would manufacture / supply UEs (devices) for this network ?". If you configure the frequency of the private to one of the licensed frequency, you would easily get a device that can work with your private network, but it is not allowed for you to use the licensed frequency without the permission from the carrier who owns the spectrum. However, if you configure the private network(eNB) to any other arbitrary frequency, it would not be easy to find any device that works with the network.

Why device manufacturere does not produce a device that can work with any arbirary frequency ? It would be mainly because of the cost for compliance / conformance testing for those devices. For each and every frequencies a device claims to support, the manufacturer should complete a huge set of conformance test and it will cost a lot. So, there wouldn't be many device manufacturer who will produce such a small set of devices for the private network.

However, if there is a certain block of a common spectrum like CBRS for Private Network, the size of device market may get large enough for device makers to jump in. This kind of market will be especially appealing for those who is manufacturing IoT devices.

 

[Note 1] If you are using CBRS band as the frequency range for your private LTE network, you may easily get the UE(e.g, SamSung Galaxy S10) supporting the frequency range (LTE band 48, specified for CBRS)

Why do we want Private Network ? (Motivation for Private Network ?)

The answer to this question might be different for different persons, but you would find some of the common advantage that many people can agreed upon from Qualcomm WhitePaper Private LTE Networks that are listed as below (I would suggest you to refer to the original whitepaper if you want to get some detailed description for each of these items).

  • Range/Link Budget
  • Spectral Efficiency/Capacity
  • Configurable QoS
  • Mobility
  • Ecosystem & Interoperability
  • High to Low Rate Scaling
  • Spectrum Options
  • Security
  • Roadmap to 5G

If you ask me about my personal motivation for adopting the private network, I would say as follows :

  • Replacement for WiFi Network : I don't think it would be cost efficient to replace your home WiFi AP with Private LTE Network. However, if it is for a whole building or multiple buildings scattered here and there, you may cover the area with much less number of LTE eNBs than the number of WLAN APs. In addition to the wide coverage of a single eNB comparing to single WLAN AP, LTE network can easily provide mobility (handover). So if your device should operate seamlessly while moving around a certain area, adopting the private LTE network can be a better option than WiFi network.
  • Getting the full control over the network : You might not have much issues of using your device subscribed to an exisiting network operators, but I don't think many people would try the commercial network for a mission critical task like industrial automation, applications requiring relatively high throughput with high reliability like operating medical equipment etc because the quality of service in the commercial network would be vary widely depending on situations. But if you use your own private network, you can allocate full network resources for your own usage and configure the quality of services as you like.
  • Gaining Network Coverage in Remote Area : If your premises are at a remote locations like a mining locaion or oil wells etc that is not so well covered by a commercial network, a Private LTE Network can be a good options to fill in the gap.
  • All the sensitive data in your premise only : If you use any device with an existing commerical network, many of the sensitive information (e.g, authentication, security related information) should leave your premise and go through the network operator and stored out-of-your premise. As another example, if your device is for collecting some sensitive information like survailence data, it also should go through the carrier network and probably go through public network before it reaches the final storage site and it may raise some security issues. With private network, you can keep all of these information / data local to your private places.
  • Cost for data plan :  This is just my thought and I haven't done any calculation on this, but this struck me leaving in a country where the cost for data faills in the category of one of the most expensive area :). I thought 'how much it would cost if I have several undreds of LTE survailence camera generating several hundreds or several Gb data in a day ?". Probably it might be more cost effective to deploy the private network and collect all those data via the private network and store them in private location.     

What has changed since this was written ?

The three questions above were asked while private cellular was still a product category rather than a specification. Two of them have since been answered by 3GPP, and the answer is worth knowing because it is not an LTE answer. The private network became a standardised concept in Release 16, and it was defined for 5GS only.

The name it was given is the Non-Public Network, and it comes in two forms. A Stand-alone Non-Public Network is independent of any operator. A Public Network Integrated NPN runs on an operator network, and access to it is restricted to a group of subscribers. One RRC information element carries both.

Following is based on 38.331 v19.3.0 (Release 19)

NPN-Identity-r16 ::=             CHOICE {
    pni-npn-r16                      SEQUENCE {
        plmn-Identity-r16                PLMN-Identity,
        cag-IdentityList-r16             SEQUENCE (SIZE (1..maxNPN-r16)) OF CAG-IdentityInfo-r16
    },
    snpn-r16                         SEQUENCE {
        plmn-Identity-r16                PLMN-Identity,
        nid-List-r16                     SEQUENCE (SIZE (1..maxNPN-r16)) OF NID-r16
    }
}

CAG-IdentityInfo-r16 ::=         SEQUENCE {
    cag-Identity-r16                 BIT STRING (SIZE (32)),
    ...                                                          -- the remaining members are not identity related
}

The CHOICE is the whole distinction. A PNI-NPN is named by a PLMN identity together with a Closed Access Group identity, which is a 32 bit string, so it is a subset of an operator network. An SNPN is named by a PLMN identity together with a Network Identifier, and the pair identifies a network that is nobody else’s.

One detail settles which technology this belongs to. The field npn-IdentityInfoList-r16 sits inside CellAccessRelatedInfo in 38.331, so a cell broadcasts it in system information. 36.331 carries none of these fields, and LTE therefore never received this framework. A private LTE network is still built the way the sections above describe, from a product rather than from a standardised identity.

Release 16 was not the end of it. Release 17 added npn-IdentityInfoList-r17 so that a secondary cell can carry its own NPN identities, and Release 18 added CAG-Config-r18. The NPN page covers the framework in full, and this page is best read as the question that framework was built to answer.

  • The private network was standardised in Release 16 : as the Non-Public Network, and only for 5GS.
  • Two forms exist and one IE carries both : NPN-Identity-r16 is a CHOICE between pni-npn and snpn.
  • SNPN is independent, PNI-NPN is not : an SNPN is identified by a PLMN identity and a NID, and a PNI-NPN by a PLMN identity and a CAG identity.
  • The identity is broadcast : npn-IdentityInfoList-r16 is inside CellAccessRelatedInfo, so a UE learns it from system information rather than from configuration.
  • LTE has no equivalent : 36.331 carries none of these fields, which is why a private LTE network stays a product question rather than a protocol one.

Reference :

YouTube

[1] Private LTE Networks

[2] Private LTE Opportunities and Applications Panel Discussion

[3] MulteFire Private LTE Demo with Nokia

[4] Qualcomm explains private LTE networks

[5] Private LTE & Multi Access Edge

[6] Private LTE over CBRS demo - Extended version