When we talk about Internet of Things most of times we are thinking in mobility. Although this is not necessary in many architectures, when we talk about some of the IoT markets like Fleet control, Connected Car, Trackers, Wearables and so on, mobility in the solution is a must.

There are different technologies to provide wireless connectivity for the IoT Market in vast areas, as LoRA or Sigfox… but because coverage, capex, inversion and operative costs Mobile Networks already rule the market and probably it continues ruling in the short future.

In fact, in order to provide IoT connectivity, what Telco operators are doing is reusing and offering their conventional 3G and 4G networks. This avoids inversion and provide connectivity to the new wave of sensors and devices which populate the Internet of Things. However this is also link to a couple of issues which are added to the perhaps the mostly known about the battery necessities of the IoT devices:

– Signalling storms – Quality of Service and/or Service Level Agreement

But, why is this an issue for a conventional mobile network? Why does this not affect to the conventional users in the same way?

First of all an IoT device has not a standarization. A mobile handset typically used by a person has a quality control which must be accomplished during its manufacturing. Sometimes it also fails and a signalling storm is created collapsing a mobile network through this kind of DDoS (it happened with an iPhone version in some networks). But, what is a signalling storm?

A signalling storm happens when lots of devices try to connect to a network, at more or less the same time. That generates a kind of DDoS attack which can collapse or affect an entire network through a problem in one of its core elements.

If thousand of IoT devices are not properly configured to avoid this (and many times it doesn’t depend of the telco) a massive simultaneous reconnection can affect in this way.

Further more, because network avaibility and global service necessities in many IoT cases of use there are roaming scenarios when a SIM Card of an operator are using the mobile network of other.

In this last case Quality of Service (QoS) and Service Level Agreements (SLA)are also a frequent gap. When the mobile networks were designed IoT was only an idea and it was not considered.

So traffic it not quality considered at End to End level. For sample: in a roaming scenario there is not difference at network level between an user who is not able to connect with Google and a telehealth device. This is an important gap.

Additionally, between operators there are not agreed quality times to resolve an incident because, in many cases, this only affects to entertaiment traffic and not industrial or critical one (except in the case of massive incidents). This is not compatible with a Connected Car solution or an International Fleet Management.

Yes, standard 4G-LTE specifies a QoS configured at APN level, but it doesn’t affect to the signalling or to the network capacity. Only to the PGW discarting policy and it not clear than in a global scenario it is going to be alligned between all the involved actors.

And here it is where standard 5G appears. Everybody knows about 5G mainly because the real time applications and low latency flows. Also de high throughputs. Everything about that is very well but the really good point of the 5G standard is try to fix the described issues. How? Response is multislicing.

Multislicing is going to be a new technology, a paradigm, which is going to permit slice the network for a kind of service. This means: book network resources to assurance the traffic quality and control at end to end of the service. Based in Cloud/Virtualization/SDN technology in case of a Signalling Storm network will be able to resize itself in order to support the avalanch. Cisco is already working in some interesting proposals in this way.

Also this would affect to all the communication chain guaranting times between nodes and avoiding possible bottlenecks in some of them.

What is not clear at this moment is how this is going to affect to a roaming scenario and how this slicing information is going to be communicated between telco operators. In this way if a Connected Car with a SIM card goes to other country through Europe how the visited network is going to know the slicing necessity and share it with the home network which owns the SIM is an issue now. It is being discussed at 3GPP level.

New vSIM technologies can help also with this, but it is not clear the prevision and if it will be valid for every operator accross the world.

Conclussion here is that yes, 5G provides an increament of the speed, a decreament of the latency and its going to permit marvellous applications. But the main point in my opinion it is going to be the first network designed 100% for Internet of Things.

Juan de la Cruz @jdelacruz_IoT

Acabo de terminar “Sobre Economía de los Datos. Riqueza 4.0”. Del catedratico Emilio Ontiveros y editado por la Fundación Telefónica. Aunque no parece el típico libro del que se hacen reseñas mantenemos esta costumbre y ponemos nuestra opinión y comentarios.

Me ha dado por acelerar el ritmo de los libros técnicos por leer, tras pasar una etapa de libros sobre filosofía. ¿Razón? En breve viajamos a Rusia a hacernos el Transiberiano.

Hay muchos ratos largos en el tren. Mis compañeros de viaje serán un grupo de amigos catalanes de los que temo me estén hablando buena parte del viaje de política,

por lo que me parece una buena iniciativa leer autores rusos mientras viajo en tren por la estepa. Próximo objetivo pues: Dovstoieiski.

Pero volviendo a nuestro libro… Si eres Data Scientist o Ingeniero de Datos no es tu libro. Continúa dándole a R, a Python o a ese nuevo Algoritmo en el que estás trabajando. El libro es sin embargo una excelente introducción explicada en un lenguaje claro y poco técnico sobre las nuevas tecnologías, la nueva sociedad que se nos viene encima, las implicaciones en varios ámbitos que tiene como el técnico, el laboral, el legal, social, privacidad…

Además, explicado en 150 páginas que se leen muy rápidamente. Como síntesis es muy buena, incluyendo la explicación técnica de que es la Inteligencia Artificial hoy en día, muy poco entendida por lo general. BigData, MachineLearning, comprensión del lenguaje, reconocimiento de patrones e Internet de las cosas.

¿Conceptos principales con los que me quedo? El OpenData, muy interesante incluyendo las iniciativas de América Latina (Diario El Clarín, mención aparte).

La demanda de profesiones relacionadas con el Dato en la Unión Europea y España (ingenieros y científicos).

La necesidad de un entorno correctamente regulado que permita tanto la legítima explotación de los datos (beneficiosa para la sociedad) como la correcta protección y transparencia de los usuarios.

Conclusión: altamente recomendado para cualquiera que desee entender y conocer este mundo a modo de introducción. Quizás de lo mejor que haya en esa línea ahora mismo.

When we think in a IoT Data adquisition system we typically have to take the decissiona about the protocol to use in order to communicate our devices with the data-collector system, as a previous step to the Big Data one.

Common discussion is about REST or MQTT using, which has sense in many cases because their light weight and flexibility… but this flexibility is also a headache because the lack of standarization and customization necessities.

MQTT typically could require a ligh JSON structure and a parser in the other side. REST needs webservice API and some development at firmware level.

http://mqtt.org/

https://www.codecademy.com/articles/what-is-rest

We propose for some cases the analysis of SNMP (Simple Network Management Protocol) for the data catching.

http://www.net-snmp.org/

The reasons of this are four:

1) It is also a light protocol. It was designed to manage wide networks without consume too much resources.

2) It is and old standard, with security in its version 3

3) There are lots of free and not-free SNMP systems in the market (IBM Webtop or BSM, HP NNMi, Nagios…)

https://www.nagios.org/

4) SNMP protocol is compatible, because its standarization, between different kind of devices. For sample with a single data catcher system (a SNMP collector) we can manage information from servers, routers and other network elementos and also IoT devices and combine all of them in a single BigData system through ETLs and cubes creators. Also you can combine incident or capacity information of the network systems with the IoT payload and extract conclusions.

Using this kind of solution, reccyling existing infrastructure, we have been able to avoid wastes in time and money. So recommendation: if you already have a SNMP collector linked to a BigData system think about reuse it also for your IoT devices always it is possible for you.

A necessity sometimes appears in a IoT voiced-based service: to collect the calls generated by the devices in a single and global centralized call center. When we say global we are talking about services with movility which can potencially be in every country in the world. For sample e-call of vehicles, fleet management services, telecare through e-health devices, pannic buttom of some infrastructures etc.

Problems with this kind of design are typically three:

– Voice international costs – Necessities of manage the device calls as a private office network, with own customization – Transformation of the voice calls through the different voice carriers and actors through the global network

Siptrunking is the solution for a wide and global office network where we are interested it works with a voice system like a local one managed by a single IP-PBX. Through SIP protocol we can avoid international voice costs and call in all our company as if we were in a same country/office. But, could this solution be applied to a car or other devices? Can we avoid in this way the effect of the different international voice carrier transformations perform in our transfers?

https://sipcity.com.au/what-is-sip-trunking/

To see this in a better way first of all we must understand how a international data roaming service works. Let’s explain this with detail:

First of all we considere a SIM card communication to establish the data channel from the device to the IP-Call Center. As we are analyzing a global scenario we must considere, as that SIM belongs to a single provider, the device which carries it out is going to be in roaming.

There are other possibilities as having SIMs of different providers (which makes complex the operation (and operational costs and the logistic processes) or the vSIM. About this last one we will talk in other post. https://www.kaspersky.com/blog/virtual-sim/11572/

In a roaming/mobile communication there always is a data channel between the SIM (or eSIM) of our device and the conventional Internet. There are two networks involved (simplifying the model): mobile one and Internet one.

SIM communicates with the border system between both (GGSN/PGW) using the APN configured in the device. Once this register is accepted we can start the data sending.

SIP protocol, as an IP based, can work perfectly in this way with the appropiate configurations in the APN.

About Quality and traffic control some bullet points to considere:

– Establish an appropiate Quality parameter through LTE protocol using de standard of the protocol https://www.wirelessdesignmag.com/article/2013/09/quality-service-over-lte-networks-part-1-3 – Establish communication close to the call center in order to avoid network delays or jumps. For that deliver the traffic in a local GGSN linked to a SBC.

This schema could look quite similar to a VoLTE communication (without IMS) and considered it part of a private voice network (instead of a international conventional one).

Main consideration is our devices and its firmware must have the capacity to manage and establish calls using SIP protocol. This is not an issue in most of hard infrastructures as cars or transport infrastructure but can be problematic in others as telecare devices where wearables have battery restrictions in some cases.

Conclusion is that when you design a deploy a global voice call IoT service the using of Siptrunking to centralize and coordinate calls of all our IoT network can be a good point to take in consideration instead of using conventional voice network.

Juan de la Cruz @jdelacruz_IoT https://www.linkedin.com/in/juan-de-la-cruz-berlanga-39856490/