ICN Research Group A. Lindgren Internet-Draft F. Ben Abdesslem Intended status: Experimental B. Ahlgren Expires: January 7, 2016 SICS O. Schelen Lulea University of Technology A. Malik Ericsson July 6, 2015 Applicability and Tradeoffs of Information-Centric Networking for Efficient IoT draft-lindgren-icnrg-efficientiot-03 Abstract This document outlines the tradeoffs involved in utilizing Information Centric Networking (ICN) for the Internet of Things (IoT) scenarios. It describes the contexts and applications where the IoT would benefit from ICN, and where a host-centric approach would be better. The requirements imposed by the heterogeneous nature of IoT networks are discussed (e.g., in terms of connectivity, power availability, computational and storage capacity). Design choices are then proposed for an IoT architecture to handle these requirements, while providing efficiency and scalability. An objective is to not require any IoT specific changes of the ICN architecture per se, but we do indicate some potential modifications of ICN that would improve efficiency and scalability for IoT and other applications. This document mainly serves as a problem statement and will not present a conclusive architecture design. It can be used as a basis for further discussion and to design architectures for the IoT. Status of this Memo This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference Lindgren, et al. Expires January 7, 2016 [Page 1] Internet-Draft ICN for Efficient IoT July 2015 material or to cite them other than as "work in progress." This Internet-Draft will expire on January 7, 2016. Copyright Notice Copyright (c) 2015 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. Lindgren, et al. Expires January 7, 2016 [Page 2] Internet-Draft ICN for Efficient IoT July 2015 Table of Contents 1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2. Advantages of using ICN for IoT . . . . . . . . . . . . . . . 5 2.1. Naming of Devices, Data and Services . . . . . . . . . . . 5 2.2. Distributed Caching . . . . . . . . . . . . . . . . . . . 5 2.3. Decoupling between Sender and Receiver . . . . . . . . . . 5 3. Design Challenges of IoT over ICN . . . . . . . . . . . . . . 6 3.1. Naming of Devices, Data and Services . . . . . . . . . . . 6 3.2. Efficiency of Distributed Caching . . . . . . . . . . . . 7 3.3. Decoupling between Sender and Receiver . . . . . . . . . . 8 4. Proposed Design Choices for IoT over ICN . . . . . . . . . . . 9 4.1. Relationship to existing Internet protocols . . . . . . . 9 4.2. Data naming, format and composition . . . . . . . . . . . 9 4.3. Immutable atomic data objects . . . . . . . . . . . . . . 10 4.4. Data naming in streams of immutable data objects . . . . . 11 4.5. The importance of time . . . . . . . . . . . . . . . . . . 12 4.6. Decoupling and roles of senders and receivers . . . . . . 13 4.7. Combination of PULL/PUSH model . . . . . . . . . . . . . . 14 4.8. Capability advertisements . . . . . . . . . . . . . . . . 15 4.9. Name-based routing vs name resolution + 1-step vs 2-step . . . . . . . . . . . . . . . . . . . . . . . . . . 16 4.10. What's naming and what's searching . . . . . . . . . . . . 16 4.11. Meta data, tagging/tracing of data . . . . . . . . . . . . 16 4.12. Handling actuators in the ICN model . . . . . . . . . . . 17 4.13. Role of constrained IoT devices as ICN nodes . . . . . . . 18 5. Security Considerations . . . . . . . . . . . . . . . . . . . 19 5.1. Retrieving trusted content from untrusted caches . . . . . 20 5.2. Enabling application-layer processing in untrusted intermediaries . . . . . . . . . . . . . . . . . . . . . . 20 5.3. Energy efficiency of cryptographic mechanisms . . . . . . 20 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 22 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 23 Lindgren, et al. Expires January 7, 2016 [Page 3] Internet-Draft ICN for Efficient IoT July 2015 1. Motivation Information Centric Networking (ICN) has been shown to efficiently meet current usage demands of computer networks, where users consume content from the network instead of communicating with specific hosts. The applications and usage of the Internet of Things (IoT) often imply information centric usage patterns, where users or devices consume IoT generated content from the network instead of communicating with specific hosts or devices. However, while the IoT shares many characteristics with typical information centric applications, it differs because of the high heterogeneity of connected devices (including sensors and actuators), the very high rate of new information being generated, and the heterogeneity in requirements from applications regarding information retrieval and dynamic actuation. Because of these differences, using an Information Centric Network to design an architecture of the IoT is often, but not always, beneficial. Depending on the context, the IoT architecture may benefit from using an ICN or a host-centric network (HCN). In practice, the right approach is a complex tradeoff that depends on the applications and usage of the IoT network. This document describes some advantages and inconveniences of using an ICN for the IoT architecture, and helps finding the right tradeoff between using an ICN or an HCN, depending on the context. In this, we explore how to represent and model IoT on top of existing ICN solutions, without requiring IoT specific functionality in the ICN. We discuss this in terms of effectiveness, efficiency and scalability. However, in some cases we do invite for discussion on tentative additions of functionality to ICN in order to make the overall IoT solution more efficient and scalable. Lindgren, et al. Expires January 7, 2016 [Page 4] Internet-Draft ICN for Efficient IoT July 2015 2. Advantages of using ICN for IoT A key concept of ICN is the ability to name data independently from the current location at which it is stored, which simplifies caching and enables decoupling of sender and receiver. Using ICN to design an architecture for IoT data potentially provides such advantages compared to using traditional host-centric networks. This section highlights general benefits that ICN could provide to IoT networks. 2.1. Naming of Devices, Data and Services The heterogeneity of both network equipment deployed and services offered by IoT networks leads to a large variety of data, services and devices. While using a traditional host-centric architecture, only devices or their network interfaces are named at the network level, leaving to the application layer the task to name data and services. In many common applications of IoT networks, data and services are the main goal, and specific communication between two devices is secondary. The network distributes content and provides a service, instead of establishing a communication link between two devices. In this context, data content and services can be provided by several devices, or group of devices, hence naming data and services is often more important than naming the devices. 2.2. Distributed Caching While caching mechanisms are already used by other types of overlay networks, IoT networks can potentially benefit even more from caching systems, because of their resource constraints. Wireless bandwidth and power supply can be limited for multiple devices sharing a communication channel, and for small mobile devices powered by batteries. In this case, avoiding unnecessary transmissions with IoT devices to retrieve and distribute IoT data to multiple places is important, and storing such content in the network can save wireless bandwidth and battery power. Moreover, as for other types of networks, applications for IoT networks requiring shorter delays can benefit from local caches to reduce delays between content request and delivery. 2.3. Decoupling between Sender and Receiver IoT devices may be mobile and face intermittent network connectivity. When specific data is requested, such data can often be delivered by ICN without any consistent direct connectivity between devices. Apart from using structured caching systems as described previously, information can also be spread by forwarding data opportunistically. Lindgren, et al. Expires January 7, 2016 [Page 5] Internet-Draft ICN for Efficient IoT July 2015 3. Design Challenges of IoT over ICN As outlined in Section 2, there are potential benefits from using ICN to implement IoT communication architectures. However, in order to obtain a scalable and efficient architecture there are some aspects of ICN that must be specifically considered in making the right design choices for IoT. In fact, using an ICN may not be beneficial in all desired sub-functions and scenarios. This section outlines some of the ICN specific challenges that must be considered and describes some of the trade offs that will be involved. We will address these challenges in our proposed design choices later in Section 4. 3.1. Naming of Devices, Data and Services The ICN approach of named data and services (i.e., device independent naming) is typically desirable when retrieving IoT data. However, data centric naming may also pose challenges. o Naming of devices: Naming devices is often important in an IoT network. The presence of actuators requires clients to act specifically on a device, e.g. to switch it on or off. Also, managing and monitoring the devices for administration purposes requires devices to have a specific name allowing to identify them uniquely. There are multiple ways to achieve device naming, even in systems that are data centric by nature. For example, in systems that are addressable or searchable based on metadata or sensor content, the device identifier can be included as a special kind of metadata or sensor reading. o Size of data/service name: In information centric applications, the size of the data is typically larger than its name. For the IoT, sensors and actuators are very common, and they can generate or use data as small as a short integer containing a temperature value, or a one-byte instruction to switch off an actuator. The name of the content for each of these pieces of data has to uniquely identify the content. For this reason, many existing naming schemes have long names that are likely to be longer than the actual data content for many types of IoT applications. Furthermore, naming schemes that have self certifying properties (e.g., by creating the name based on a hash of the content), suffer from the problem that the object can only be requested when the object has been created and the content is already known, thus requiring some form of indexing service. While this is an acceptable overhead for larger data objects, it is infeasible for use when the object size is on the order of a few bytes. Lindgren, et al. Expires January 7, 2016 [Page 6] Internet-Draft ICN for Efficient IoT July 2015 o Hash-based content name: Hash algorithms are commonly used to name content in order to verify that the content is the one requested. This is only possible in contexts where the requested object is already existing, and where there is a directory service to look up names. This approach is suitable for systems with large data objects where it is important to verify the content. o Metadata-based content name: Relying on metadata allows to generate a name for an object before it is created. However this mechanism requires metadata matching semantics. o Naming of services: Similarly to naming of devices or data, services can be referred to with a unique identifier, provided by a specific device or by someone assigned by a central authority as the service provider. It can however also be a service provided by anyone meeting some certain metadata conditions. Example of services include content retrieval, that takes a content name/ description as input and returns the value of that content, and actuation, that takes an actuation command as input and possibly returns a status code afterwards. 3.2. Efficiency of Distributed Caching Distributed caching is a key opportunity with ICN. However, an IoT framework must be carefully designed to reap the maximum benefits of ICN caching. When content popularity is heterogeneous, some content is often requested repeatedly. In that case, the network can benefit from caching. Another case where caching would be beneficial is when devices with low duty cycle are present in the network and when access to the cloud infrastructure is limited. However, using distributed caching mechanisms in the network is not useful when each object is only requested at most once, as a cache hit can only occur for the second request and later. It may also be less useful to have caches distributed throughout ICN nodes in cases when there are overlays of distributed repositories, e.g., a cloud or a Content Distribution Network (CDN), from which all clients can retrieve the data. Using ICN to retrieve data from such services is beneficial, but in case of dense occurrence of overlay CDN servers the additional benefit of caching in ICN nodes would be lower. Another example is when the name of the data has a different meaning depending on the context, or if the name refers to an object with variable content/state. For example, when the last value for a sensor reading is requested, the returned data should change every time the sensor reading is updated. In that case, ICN caching may increase the risk that cache inconsistencies result in old data being returned. Lindgren, et al. Expires January 7, 2016 [Page 7] Internet-Draft ICN for Efficient IoT July 2015 3.3. Decoupling between Sender and Receiver Decoupling the sender and receiver is useful mechanism offered by the ICN approach, especially for content retrieval with duty cycling devices or devices with intermittent connectivity. However, in order to efficiently retrieve data it must be possible for requestors (receivers) to easily deduce the name of the data to request, without any direct contact with the responder (sender). Nevertheless, de-coupling is a challenge when authentication is needed for management and actuation, or when real-time interaction between devices is necessary. Solutions for object security supporting decoupled authentication (e.g., similar to signing by proxy), and solutions for pushing data to decoupled entities must be explored. Lindgren, et al. Expires January 7, 2016 [Page 8] Internet-Draft ICN for Efficient IoT July 2015 4. Proposed Design Choices for IoT over ICN This section describes some fundamental design choices and trade-offs to allow for effective, efficient and scalable handling of IoT data in an ICN network. An objective with these choices is to facilitate that an ICN network can be used without requiring additions of IoT application specific functionality in the ICN network. However, in some cases we do invite for discussion on tentative additions of functionality to ICN in order to make the overall IoT solution more efficient and scalable. 4.1. Relationship to existing Internet protocols IoT devices can have a role as content generators (e.g., sensors) in where an ICN paradigm should be effective for data retrieval and dissemination. However, IoT devices may also have roles as actuators in which such devices shall be accessed for control purposes. The use of an ICN network may be less natural when actuation and control of specific devices is the key objective. As ICN networks are likely to coexist with existing Internet protocols in most situation, often being deployed as overlay networks, we will consider that there may be situations where a host centric addressing is more suitable for IoT. Thus, to facilitate support of IoT for both data generation and control/actuation, we assume that ICN routing should therefore work in concert with existing Internet protocols. However, we will also investigate the possibility of utilizing ICN network primitives for actuation as well to see what the tradeoffs are, as can be seen in Section 4.12. 4.2. Data naming, format and composition The data served by ICN may be aggregated from smaller components. Although IoT data components in many cases are small and simple, a general challenge in defining ICN applications is to decide how to compose (i.e. group) the data so that it can be effectively named and requested. Requesting partial data inside a composition may become a challenge. Indeed, if data is composed and sub components are requested, which are not directly namable by the requestor, finding such a subset will resemble a database query which may require processing to resolve. The ICN network should not have to support such complexity. A design choice regarding IoT data is therefore to keep the ICN network free from supporting any advanced queries and instead only support directly addressable (i.e., named) data objects. Any advanced composition (hierarchical, graph-based, hyperlink, etc.) of IoT data, and related searching for sub-components, would be handled in servers/endpoints instead of inside the ICN network. The issue of Lindgren, et al. Expires January 7, 2016 [Page 9] Internet-Draft ICN for Efficient IoT July 2015 IoT data structure and searching at that higher level is for further study. For effective ICN interoperability, only the structure of the atomically addressable data objects must be agreed and mapped to the underlying ICN naming scheme. This is to avoid making new requirements on the ICN and to make sure that the need for computation is kept low in the ICN network, essentially limiting it to deciding whether there is a cache hit or not. There are some considerations following from this design choice. First, the size of the directly addressable objects could be kept fairly small to avoid that unwanted data is pulled over resource constrained networks and cached in the ICN network (resulting in better resource utilization, better localization of desired data, and ultimately better scalability). There is however a tradeoff in that smaller data objects results in a larger naming overhead. Second, this approach means that a flat ICN address space would be sufficient, but for practical reasons a hierarchical address space may add some benefits. In any case, there is flexibility in using different addressing schemes depending on what is supported by the existing ICN framework. 4.3. Immutable atomic data objects The number of IoT devices as well as the amount of data produced by these devices may potentially be very large, and the data may be spread over very large ICN networks. The potential problem of cache inconsistencies in an ICN network may therefore be large if we allow for data to be mutable objects. To support scalability and horizontal distribution it is essential to define data properties that facilitate independency and consistency, while minimizing the need for dynamic global synchronization. A key design choice is therefore to mandate that IoT only uses immutable atomic data objects. This supports large scale distribution by ensuring that there is no stale data in the ICN domain. A cache hit is always a clean hit. A trade-off from this is that dynamic data must be modeled as a stream of immutable data objects, potentially consuming more resources. However, this challenge can be resolved by smart caching strategies where old data is dropped. There is however some practicalities to consider. Devices, including IoT devices, are restarted now and then. They might in this process loose their state, including what name they used for a particular data value. So in practise it will be hard to implement a strong assumption on immutable data. We therefore likely must be prepared to handle the occasional exception to this rule. Lindgren, et al. Expires January 7, 2016 [Page 10] Internet-Draft ICN for Efficient IoT July 2015 4.4. Data naming in streams of immutable data objects Many IoT devices produce new sensor readings or other data values at regular intervals or on demand. With the design choice on immutable atomic data objects, there is a need to model the resulting stream of sensor readings with a stream of immutable data objects in the ICN domain. The need in this situation is very similar, if not identical, to video streaming, where video frames or chunks are immutable data objects in a video stream. A key advantage of modelling IoT data as a stream of immutable data objects is that ICN caches will not contain any stale data w r t a given name. However, since new data objects (with new names) representing different versions of a sensor reading may be emitted frequently, there must be a way to tell the different versions apart. To support immutable streamed data efficiently, while adhering to the expected naming schemes of ICN, we recommend that names of data objects include a sequence number. When data can be named with sequence number, any request may or may not include such a sequence number. If no number is included in the request, the nearest cache hit will result in a response. If a sequence number is included in the request, only an exact cache match will result in a response. A client that wants the "latest" reading can according to our previously mentioned design choice, in Section 4.2, not ask the ICN network such a high level query, instead it must ask for the specific (version of) information. To avoid complicated searching in the ICN nodes, there is thus no way to explicitly ask the network for the "latest" reading, or any other "range" of sequence numbers. Should a client want the latest reading from a sensor, one method for this is to make a subscription for the pushed stream of data, as described in Section 4.7, provided that the particular ICN architecture supports this interaction model. The confirmation of that subscription can contain the latest reading, and then obviously the normal stream will be received. The reason for including the latest reading in the response is to immediately provide the "state" of sensors that generate new data infrequently. Another method to obtain the latest reading, or a particular reading in the past, from a sensor is to perform adaptive probing, for example by binary interval reduction. If a requested sequence number does not (yet) exist, there will be a negative answer from the ICN. This method is preferably combined with application knowledge, for example, in the form of capability advertisements as described in Section 4.8 that enable the client to better predict the sequence number to request. The client that always wants the latest value could also dynamically tune its requests for the next data value to Lindgren, et al. Expires January 7, 2016 [Page 11] Internet-Draft ICN for Efficient IoT July 2015 the frequency of the publisher in order to minimise the latency. The fact that non-existing data is asked for would however potentially pose an overload threat to the ICN since each request of non-existing data could result in cache misses that ripple through all the way to the source, which has to respond that the data doesn't exist. It may therefore be beneficial with negative caching. Serving requests for non-existing data is however a generic challenge to ICN (not specifically to IoT) to be resolved. There is a third method "in between" the above two. If requests for a not yet existing data object can be held for a short time until the data object is actually available, instead of immediately returning "not found", these requests act as one-time subscriptions. Provided that request aggregation is being used, this mechanism would be efficient and latency-minimising, and at the same time would not require persistent subscription state. The support for sequence numbers depends on the particular flavor of ICN. The naming scheme of CCN/NDN may here provide an advantage. It is for further study whether it is possible to use ICNs that do not support sequence numbers as part of naming (e.g., by clever use of metadata, namespace, and search functionality) and what the trade- offs would be. Two issues for further study are the size of the sequence number space and gaps in the sequence numbers. Must sequence number wraparound be handled, or is it possible to require a large enough sequence number space? Wraparound means an exception to the assumption on immutable objects. Gaps in the sequence number space might result in inefficiencies in some of the above methods, or, if the gaps are large, making them unfeasible. Yet, it might not always be possible to guarantee that there are no gaps. 4.5. The importance of time Time is almost always a very important property of IoT data, and especially so for data that change over time. When modeling dynamic IoT data with a stream of immutable data values, it is often the case that a certain IoT data value is a sensor reading at a particular point in time, and the next value in the stream is the next reading in time. Thus, dynamic data is in this case dynamic over time, with well defined (immutable) values for particular points in time. We therefore argue that it is important to find a way to represent these time-related streams of immutable data values in ICN. It should be possible to request a data value from a certain time, and to infer/find the name (sequence number) of the latest, most current, data value. The question is whether or not the stream sequence Lindgren, et al. Expires January 7, 2016 [Page 12] Internet-Draft ICN for Efficient IoT July 2015 numbers are sufficient to support time. If not, the ICN system needs to be extended with explicit support for time, something we want to avoid. In general, the methods outlined in the previous section are applicable for finding an IoT data value from a particular point in time, including the latest. What is missing is the mapping between sequence number and time. One possibility could be to use sequence numbers that directly correspond to time, for instance, the Unix (POSIX) time in form of seconds since January 1st, 1970. This would however both limit the time resolution to seconds, and also result in large gaps in the sequence numbers, something that can be problematic, as discussed in the previous section. There are several other methods for finding readings from a certain time, or the latest reading, for example through a high level request from a server/endpoint, or by using a naming scheme where the name can be directly inferred, e.g., if an IoT device has advertised under which conditions it produces data and how it is named. To represent absolute time so that it can be directly inferred, one method is that the producer of data in its capability advertisements (Section 4.8) provide a mapping function between sequence number and time. Thereby also readings on the time axis are immutable while it is still possible to efficiently find the latest reading, as described in Section 4.4. It should be noted that sequence numbers then may have gaps in order to cater for triggered non periodic data, etc. Another method is to include meta data with information on absolute time. Using this mapping scheme data from the current second can be efficiently requested (provided that clock synchronisation is accurate enough, which is out of scope of this document). We also note that time is also important for other applications, in particular for live streaming video. Live video also produces a time-related stream of immutable objects, and would in the same way benefit from such support in the ICN service. 4.6. Decoupling and roles of senders and receivers Since ICN networks essentially support a request/response model of interaction, we denote the receivers of information as requestors, and the senders of information as responders. The ICN network in itself provides decoupling of requestors and responders. It is an important feature of the ICN that it will allow responders (e.g., IoT devices) to be occasionally unreachable (e.g., due to intermittent connectivity, low battery level, duty cycling). Another advantage is that caching in the ICN will ensure that data objects are normally Lindgren, et al. Expires January 7, 2016 [Page 13] Internet-Draft ICN for Efficient IoT July 2015 delivered only once from the IoT devices, independently of the number of immediate requestors. Note however, that the ICN does not (and should not) provide any transformation or aggregation of data. The IoT dissemination architecture should therefore allow for any number of intermediate processing nodes. An intermediate node will be an endpoint in the ICN network that can act as both requestor and responder. Such a node may perform aggregation, filtering, selection, etc. The instantiation of such nodes may for example form a directed (acyclic) graph between ultimate responders (IoT devices) and ultimate requestors (the final applications). It is for further study how to define such an architecture. It is a design choice to keep the IoT dissemination and aggregation functionality outside of the ICN domain. That architecture would be an overlay that may have intricate structure, and put the ICN usage in a new context, where content from ultimate requestors to ultimate responders may go through many IoT processing nodes that collect, process and re-publish data through an ICN for various purposes. 4.7. Combination of PULL/PUSH model A critical decision regarding IoT data is whether to use a PULL model, a PUSH model, or both. In this document, we define a PULL model as a system where data is only sent when explicitly requested, while a PUSH model indicate that data transmission is initiated by the source based on some trigger (either periodic, for each new object, or based on some condition on the generated data). There are some intrinsic trade offs between these models. The PULL model is for example resource efficient when there is an abundant amount of IoT information, potentially redundant from many devices, and the clients only occasionally or partially are interested in the information. The PUSH model is for example efficient when there is real-time information and the clients are interested in all information from specific devices all the time. A design decision in the IoT domain is to support both PULL and PUSH. The base model should be PULL, since this is the native mode of ICN, meaning that requestors must always start by sending a request. If the request is for some specific data, it can be resolved by returning the data (if it exists). The pull model can be supported efficiently and scalably by an ICN network. A challenge with the pull model is that it may be inefficient for retrieving new data that occur sporadically or based on specific conditions. Our proposal for an IoT framework is therefore that there must be support for efficiently retrieving such triggered Lindgren, et al. Expires January 7, 2016 [Page 14] Internet-Draft ICN for Efficient IoT July 2015 information, without having to poll for it through the ICN. Our proposal is that a request can also include triggers, which means that data will be returned (pushed) when triggers are fulfilled, which may be immediately, or in the future at one or several occasions. This can be used to select alarm conditions, to request continuous or periodic push, etc. The trigger conditions could be set by the requestor, or be pre-defined by the responder. The former would be more flexible but also have performance/scalability issues since the number of trigger conditions and consequent data generation would depend on a potential large number of requestors. The latter is more scalable since there will be a predefined and finite number of trigger conditions (as defined in capability advertisements). Our recommended choice, at least for the initial phase, is to go for a simple and scalable solution and therefore adopt the model where available trigger conditions are defined and advertised by the responder. The ICN would be apt for supporting such capability advertisements, given that they are fairly static. With this, there is no requirement raised on ICNs supporting data push, but we recommend to have a discussion on whether an ICN network can or should provide an option to effectively support a push model of data. Such support could make real-time IoT data dissemination more efficient and scalable as previously mentioned in Section 4.3. However, since we assume that the ICN works with existing IP protocols, such functionality can be provided without ICN, by using traditional unicast or multicast communication. We finally note that an ICN supported push service model would make the ICN network more like a publish/subscribe system. 4.8. Capability advertisements Capability advertisements and discovery can be used by requestors to discover which data is available and/or to which responders to connect to. In a deployment with large numbers of responders, the functionality of automatic advertisement and discovery becomes a critical factor to support scaling. Responders should advertise their methods (inputs, outputs, parameters, triggers, etc) and provide relevant metadata in the responses as advertised. Such capability advertisements should be conservative with resources, which suggests that new advertisements should be posted with reasonably low frequency. This implies that an ICN network can be used for providing capability advertisements. The advertisements should be provided as a stream of immutable objects, or alternatively the IoT system should be tolerant to stale caches. Should there be a need real-time awareness of dynamic changes, a subscription/push model of capability advertisements could be used as earlier described in Section 4.7. Lindgren, et al. Expires January 7, 2016 [Page 15] Internet-Draft ICN for Efficient IoT July 2015 4.9. Name-based routing vs name resolution + 1-step vs 2-step As described in Section 4.2, the IoT framework should be defined so that new functionality in the ICN is not needed. For data that is frequently generated and regenerated, it makes sense to keep simple structures and provide directly inferable naming/addressing of data objects, so that requestors can directly address the data. For more complex data, such as pre-processed, aggregated and structured data a two-step resolution model is recommended. The IoT devices can provide a higher level resolution based on for example queries and searching, resulting in a number of concrete directly addressable ICN objects. This is similar to what web servers do when they return URLs that requestors can use, but in this case it is named content that is returned. Consequently, the IoT framework should have no requirement that the ICN network itself should support 2-step addressing (although such 2-step methods may exist in some ICNs) 4.10. What's naming and what's searching As described in Section 4.2, the IoT framework should be defined so that no new functionality is required in the ICN for searching data or subcomponents of data. The ICN network supports just naming of atomic data objects, while any searching is provided by the IoT framework, which in itself may be constituted by a highly distributed set of nodes that provide processing, analysis and aggregation of IoT data. 4.11. Meta data, tagging/tracing of data IoT data may be tagged with metadata to tell where it originates from. Tagging is made at the level above the ICN network and may for example be a list of strings. It can be added/changed by the originating node (or a node that assigns the originating ID), and added/changed/deleted by any node that processes the data. The tag can in some cases be used to trace data back to origins. In some cases it makes no sense to request or transmit metadata. For efficiency reasons the ICN network should have support for optional delivery of metadata. This is to be conservative with scarce resources, for example when a wireless node requests data which is cached in the ICN network, it would be beneficial if the requestor could tell that it is desirable to not receive any metadata. There should be a discussion whether there should be just one, or more than one, piece of optional information in ICN content to be future proof. Lindgren, et al. Expires January 7, 2016 [Page 16] Internet-Draft ICN for Efficient IoT July 2015 4.12. Handling actuators in the ICN model If actuators should be controlled using the ICN communication model, we need to map the functionality of the actuator to named data and/or the requesting of named data. We see two main models with some variants as described in the following paragraphs. In the first model, the state of the actuator is represented by a stream of immutable named data objects. The actuator periodically requests its new state using the name of its designated state object. There then has to be a producer of that state data that responds with the current state. When the actuator receives the response, it sets that new state, invoking its actuation function. Authentication of the producer of the state is important, but as this corresponds directly to publisher and data object authenticity that are fundamental in the ICN model, there are no additional requirements for the IoT domain. A variant of this first model is that a requester first requests the state of the actuator. The requester supplies additional information with the request including the name of the new state data it will produce. The actuator responds with its state, and then requests its new state using the name that was supplied with the additional information in the first request. This variant enables low latency without high frequency polling. In the second model, the actuator invokes its actuation function as a side-effect of receiving a particular request. There are several plausible variants. The new state could be encoded in the name of the requested data in the request, or could be supplied as additional information with the request. Regardless, the actuator acts on the new state information as a side effect, and responds with data, possibly its state, to the requester. The security issues are potentially more difficult with this model, since in the ICN model, anyone could make the request. Access control and/or requester authentication are therefore required. To reap the advantages of caching, it should be possible to cache the state of the actuator in both the aforementioned models. However, we think that caching is not as relevant for actuation as it is for other IoT use cases, and can furthermore even be quite problematic. The first model less so, since the actuator can make sure that its state is arbitrarily fresh with the polling method described in Section 4.4. In other words, the latency until actuation happens can be bounded. The variant of the first model and the second model have larger issues. With caching, it is hard for a requester to make sure that its request actually reaches the actuator, and thus, it is hard to bound the actuation latency. Some caching directive might be Lindgren, et al. Expires January 7, 2016 [Page 17] Internet-Draft ICN for Efficient IoT July 2015 needed in this case for reliable functionality. 4.13. Role of constrained IoT devices as ICN nodes Typical ICN nodes such as routers and gateways are deemed to be rich in resources like energy, processing, bandwidth and storage. IoT devices, on the other hand, are quite constrained in such resources. It is also worth noticing that some resources are more crucial than others. In most cases energy, processing and bandwidth are quite expensive for constrained IoT devices. In contrast, storage has shown a considerably rapid decreasing trend in prices over the past few years. There is reason to believe that the memory needed for IoT devices to act as servers of their data will not be prohibitive and that the data centric role of the devices may be elevated by information-centric networks. However, it is questionable whether IoT devices also should provide caching for data produced by other IoT devices. In ad-hoc networks this may be desirable, but often there is a desire for wireless nodes to minimize communication by handling only data of their own concern. Our design decision in this regard is that we logically separate IoT server functionality (such as sensing and transmitting IoT data) and ICN functionality (such as routing and caching data generated by other devices). A resource constrained device may choose to only implement IoT functionality and act as a server to the ICN, i.e., not act as intermediate ICN node. However, since storage is getting cheaper, IoT devices should be able to cache their own content and, in essence, act as sources to ICN. Lindgren, et al. Expires January 7, 2016 [Page 18] Internet-Draft ICN for Efficient IoT July 2015 5. Security Considerations The ICN paradigm is information-centric as opposed to state-of-the- art host-centric internet. Besides aspects like naming, content retrieval and caching this also has security implications. ICN advocates the model of trust in content rather than trust in network hosts. This brings in the concept of Object Security which is contrary to session-based security mechanisms such as TLS/DTLS prevalent in the current host-centric internet. Object Security is based on the idea of securing information objects unlike session-based security mechanisms which secure the communication channel between a pair of nodes. This reinforces an inherent characteristic of ICN networks i.e. to decouple senders and receivers. In the context of IoT, the Object Security model has several concrete advantages. As discussed earlier in Section 2.1, in many IoT applications data and services are the main goal and specific communication between two devices is secondary. Therefore it makes more sense to secure IoT objects instead of securing the session between communicating endpoints. It is important that while security mechanisms complement the ICN architecture in a coherent fashion, they do so without laying down any strict requirements or constraints. Therefore, the decision of what security mechanisms are employed should be handled at a layer above ICN, in this case within the IoT framework. However, the ICN layer should not be completely oblivious of Object Security. At this point it is important to distinguish between the different aspects of Object Security i.e. integrity, authenticity and confidentiality. ICN provides data integrity through Name-Data Integrity i.e. the guarantee that the given data corresponds to the name with which it was addressed. Typical ICN protocols provide Name-Data integrity using various schemes such as hash-based names and signatures. Signature-based schemes additionally provide data authenticity. Otherwise data authenticity should be provided in layers above the ICN layer. Data confidentiality should also be handled above the ICN layer. This facilitates flexibility and allows IoT applications more freedom to decide which encryption scheme suits them best (session- based encryption, object-based encryption or a hybrid). Though the idea of Object Security is very much in line with the ICN concept, there can still be some use cases where Object Security does not add much e.g. a Pub/Sub interaction where a client is expected to interact more or less with the same server node (a session-based security protocol should suffice here) or use cases where application layer headers should also be secured (which can be achieved by TLS/ DTLS). We, therefore, effectively imply that there is no need to modify typical ICN standards to accommodate Object Security in its Lindgren, et al. Expires January 7, 2016 [Page 19] Internet-Draft ICN for Efficient IoT July 2015 entirety. The following sub-sections discuss some security advantages of using ICN and Object Security in IoT applications. 5.1. Retrieving trusted content from untrusted caches When functioning in an ICN network, an IoT client is expected to rely on the network to deliver the requested content in an optimal fashion without concerning itself with where the content actually lies. This could potentially mean that each individual object within a stream of immutable objects is retrieved from a different source. Having a trust relationship with each of these different sources is not realistic. This gives rise to the need of retrieving trusted content from untrusted nodes/caches in an ICN network. Through Name-Data Integrity, ICN automatically guarantees data integrity to the requester regardless of the source from where it is delivered. Additionally, Object-based signatures and encryption are ideal in such use cases because it relieves an IoT client application from the hassle of having to establish trust with each node that can potentially cache an IoT object. This also means that a requesting client can make use of more caches in the network, hence resulting in better throughput and latency. 5.2. Enabling application-layer processing in untrusted intermediaries Securing content at the object level provides greater granularity and hence more control. An ICN data object may comprise of several distinct application-layer objects e.g. XML, JSON objects. An example of this is an ICN object that corresponds to all the sensor readings in a certain time interval where each sensor reading is a JSON object. Using Object-based encryption to provide data confidentiality allows for the possibility to encrypt a subset of these application-layer objects while leaving others unencrypted and available for processing in untrusted intermediary nodes (e.g. proxies and caches). With this approach, the IoT application has more control of the parts of data it wants to make public and the parts of data it wants to keep confidential and visible only to peers with the right cryptographic keys. 5.3. Energy efficiency of cryptographic mechanisms Session-based security protocols rely on the exchange of several messages before a secure session is established between a pair of nodes. Use of such protocols in constrained IoT devices can have serious consequences in terms of power efficiency because in most cases transmission and reception of messages is more costly than the cryptographic operations. This is especially true for wireless Lindgren, et al. Expires January 7, 2016 [Page 20] Internet-Draft ICN for Efficient IoT July 2015 devices. The problem is amplified proportionally with the number of nodes the constrained device has to interact with because a secure session would have to be established with every node. If the constrained device is acting as a consumer of data this would mean setting up secure sessions with every caching node that the device retrieves data from. When acting as a producer of data the constrained device would have to setup secure sessions with all the consumers. The Object Security model eliminates this problem because the content is readily available in a secure state in the network. IoT devices producing data can secure it w.r.t. all the intended consumers and start transmitting it right away. Lindgren, et al. Expires January 7, 2016 [Page 21] Internet-Draft ICN for Efficient IoT July 2015 6. Acknowledgements The work behind and the writing of this document are in part supported by the activity `14010 Efficient IoT Content' within EIT Digital (formerly EIT ICT labs). Lindgren, et al. Expires January 7, 2016 [Page 22] Internet-Draft ICN for Efficient IoT July 2015 Authors' Addresses Anders F. Lindgren SICS Swedish ICT Box 1263 Kista SE-164 29 SE Phone: +46707177269 Email: andersl@sics.se URI: http://www.sics.se/~andersl Fehmi Ben Abdesslem SICS Swedish ICT Box 1263 Kista SE-164 29 SE Phone: +46705470642 Email: fehmi@sics.se URI: http://www.sics.se/~fehmi Bengt Ahlgren SICS Swedish ICT Box 1263 Kista SE-164 29 SE Phone: +46703141562 Email: bengta@sics.se URI: http://www.sics.se/people/bengt-ahlgren Olov Schelen Lulea University of Technology Lulea SE-971 87 SE Phone: Email: olov.schelen@ltu.se URI: Lindgren, et al. Expires January 7, 2016 [Page 23] Internet-Draft ICN for Efficient IoT July 2015 Adeel Mohammad Malik Ericsson Kista SE-164 80 SE Phone: +46725074492 Email: adeel.mohammad.malik@ericsson.com URI: Lindgren, et al. Expires January 7, 2016 [Page 24]