1 Introduction
2 Formulations of the Concepts for Novel IoT Communication Models
2.1 Conceptualisations of a Case for Liberalising IoT Deployment
2.1.1 Proactive Nature of Data Flows in IoT Communication
2.1.2 Different Fundamentals of Search and Discovery of IoT Devices, Services and Data
-
Physical location: e.g. when in a city area such as a street, a search can include the actual physical location as the main search criteria. The physical location can also include geographical coordinates/tags, regions, streets, building etc. Currently, users know the service in advance by knowing the web address/apps of the IoT provider in an area. Running a search with a city’s region such as a street using the traditional Internet search engine would be futile or require a skilful or lengthy searching to discover the location-related IoT service(s) of the actual IoT provider.
-
Timely dimension: what is being searched for often has a real meaning if time is known or recorded.
-
Ownerships: knowing the responsible company, public body or individual (sometime even devices’ manufacturer) that own the IoT devices or services and certify the data, would make the search and data fetching more meaningful. Quite importantly, it would define the trust and data integrity framework.
-
Connectivity: a defining item of the search can include network or access location, such a seeking specific (cellular) network operators or wireless access points. This option also applies to addresses (e.g. IP address or subnet5) as connectivity identifiers.
-
Data Name, Attribute, Status and Query Logic: running a generic search such as “temperature” or “traffic congestion”, a URI name or path/query name segment (more on naming is given in Sect. 3), or applying a simple query logic (e.g. seek only air pollution values from a public provider in a city location) would render a targeted and relevant resolving of the information. In addition, an open search could return some of the previous items such as the physical location, time, ownership, cell ID etc. In other words, it can facilitate the actual discovery of IoT devices and their data relative to a user’s location.
2.1.3 Emerging Novel Networking and Routing Models
-
On-demand and flexible routing installations: This is being enabled by Software Defined Networking (SDN) solutions [22] via abstracted separations of control and data planes in network infrastructures. A control entity (control plane) can install routing commands in programmable routers (data plane) upon a flow/session communication request. A communication request in IoT can be search and discovery step (as an “IP-less” layer2/3 message), hence, network’s control can point to data location6 or install routing pointers/deliver the actual data. In addition, data flows in IoT are often localised, sporadic, bursty and small. SDN solutions can surpass the conventional routing rules of today’s networks and allow for (localised) networking framework that accommodates for the specific nature of IoT communications.
-
Routing based on the context of the search: The pivotal new thinking is embodied in the Named-Data Networking (NDN) [23] proposal that follows an ongoing line of research under Information Centric Networking [24] for traditional Internet type communications. To summarize, data that is sought is specified as the Interest search packet by the receiver/seeker (e.g. data is identified similar to URLs) and such Interest travels “upwards” towards the data source or intermediate storing routers as a search-and-path-establishment packet. When the data content is found at an intermediate router, it follows back the search path. The concept has initially been recognised as very fitting to IoT environments [25] and has already been proposed and trialled as a solution for IoT search and collections at scoped campus-type environments [26, 27], automation scenarios in buildings [32], direct intercommunications between vehicles [20] etc. In these cases the search Interest packet specifying the assigned data name is a preset data structure that specifies the data type (e.g. temperature, humidity, noise, voltage, on/off indications for environmental conditions or actuation states…) and the named location (e.g. a room in building [27]). Recently, a summary of NDN potentials and solutions for IoT was surveyed and grouped as ranges of issues and challenges [28]. In parallel, there is a notable novel proposal for ICN/NDN-based IoT solution [29] using a slightly different search and routing model than the mentioned NDN for IoT solutions but following the same conceptual foundations. Rather than performing a name-based search from the very initial search step, in [29] the search first locates the root CoAP-based Resource Directory (RD) as specified in [30] using IP address/URL of the RD (i.e. via DNS or pre-configuration with RD’s IP address). From then on, the search can continue to other RDs located at nearby network elements (i.e. gateways, access points, routers) using NDN principles. Interestingly, the solution in [29] does not directly use a specific NDN-type names for IoT (e.g./ndn/kcl.edu/strand_building/1st_floor/room_17/temperature) for each IoT resource but uses a more scalable and manageable generic and opportunistic search process using attributes of IoT data (e.g. “traffic”) and following the format of IoT data annotations used by CoAP [11] and specified as CORE Link Formats [31]. If an immediate match is not found, the RD distributes the attribute-based search to all network elements acting as RDs before match(es) are found. While these solutions are already on the path of offering realization of some of the concepts proposed in this paper, in a liberalised IoT scenario, search and discovery would be much broader and liberalised in terms of the information that describes the search (as explained in bullet point 2) of this section and in Sect. 3) and would be done on a larger and open scale by being facilitated by the whole communication setup as in city environments. Hence, the horizontal networking concept is highlighted to indicate the shift towards “horizontal” control and data flow directions.
3 Towards Enablers for Liberalised IoT Deployment
3.1 Envisaging Scenarios
3.2 Drafting Enablers
3.2.1 Identifying IoTs
3.2.2 IoT Data Distribution
-
The search and discovery process ought to be opportunistic and near-match: Not all IoT devices need to be identified with a whole range of possible identification fields. Hence, a search would be resolved with a near-match. E.g. a search for parking meter occupancy in a specific street using “the street” as the search item would return the data matching the search. There might be parking meters that don’t have “the street” in their identification fields. Similarly running a search on a network cell as the search field might only return the values for IoT devices that have the “the network cell” in the identification field (and are connected to the cell’s operator). Such optional relaxation of the identification fields would make the whole system grow in opportunistic but liberalized manner that can be coherently organized or standardized in its mature implementation stages. It would surpass the scalability concerns of some NDN for IoT solutions that are based on exact name searches when generally distributed in networks (name scaling can happen only within scoped regions, e.g. buildings) and matches the “attribute” based opportunistic search as in [29].
-
Locations and ownership of databases holding IoT identification: Envisaging the locations of databases breaks the traditional layering notions. Also, data can be localized and not repeated in all databases, the search location can determine the scope and depth of the response. IoT identification data primarily ought to be timely and give the closest to real time readings of the values on the ground, or in other words, the latest. Such a requirement projects differently depending on the type of IoT devices and period of their reporting of data. Location-wise, the databases storing dynamic and timely IoT identification data can span from commercial databases in an IoT eco-system (i.e. a smart city company) to public databases storing utility data, transport, environmental data, citizen’s IoT devices etc. Similar to NDN concepts, IoT meta-data can be stored in the networking infrastructure as caches, i.e. a concept of a storing (access) router [23]. E.g. a router or access point would usually be part of a network operator or wireless access provider [29]. For cellular and traditional networks, the data storage capability in some of the transit components of the infrastructure would require similar changes as proposed in NDN, even if the storage is minimal and reduced to meta-data from the local IoT devices. One example of a wireless access provider is the access and routing infrastructure used for smart grid solution by Cisco, were storage extensions can be provided by proprietary components in the network (e.g. based on Cisco’s CGR routers). Finally, the straightforward location for localized data directories keeping IoT identifications is gateways. This is already provisioned for CoAP implementation of resource directories as mentioned previously [30] regarding the action of resource discovery. Some examples of data locations are depicted in Fig. 3. Solutions in this paper propose more comprehensive search-related data such as physical locations, network access identifications etc. In fact, these can be a pre-step to discovery of gateways. This goes back to what is mentioned above that a near-match search returns a response that is subject to the search, a search for data might return an address of a gateway, from where the CoAP research discovery can be applied (as shown in Fig. 2) for discovering the actual resources.×