Dirk Kutscher

Personal web page

Archive for the ‘RIOT’ tag

Information-Centric Long-Range Networking: Re-Imagining LoRaWAN

without comments

LoRaWAN is a popular low-power long-range communication system for IoT that is suitable for single-site deployments as well as for larger networks. It consists of LoRa, a PHY layer that allows for radio communication between 2 and 14 km, and higher-layer protocols mainly to upload IoT data to a serverbased infrastructure. These characteristics make LoRaWAN a promising option for many urban and rural IoT scenarios.

The LoRaWAN network design incurs, however, four notable shortcomings:

  1. LoRaWAN is heavily optimized towards retrieving data from constrained Nodes. Sending data to Nodes is expensive and involves significant latencies. Many networks such as the popular community The Things Network (TTN) thus deprecate sending data to Nodes above a very low message rate, making LoRaWAN unsuitable for most control scenarios.
  2. LoRaWAN has not been designed with the objective to provide a platform for Internet protocols. It is possible to use IP and adaptation layers on top of LoRaWAN, albeit very inefficiently.
  3. The whole LoRaWAN system is a vertically integrated stack that leads to inflexible system designs and inefficiencies. For example, all communication is channeled through LoRaWAN Gateways as well as Application- and Network Servers that interconnect with applications.
  4. The centralization and lock-in into vertical protocol stacks challenge data sharing (between users) and the creation of distributed applications (across LoRa island and the Internet).

A new LoraWAN architecture based on DSME and ICN

In our IFIP Networking 2022 paper "Long-Range ICN for the IoT: Exploring a LoRa System Design", Peter Kietzmann, José Alamos, Thomas C. Schmidt, Matthias Wählisch, and myself aim for a better integration of the LoRa-based Internet of Things into the remaining Internet. We base our system design on the following four requirements:

  1. enabling LoRa networks and Nodes in these networks to communicate directly with hosts on the Internet;
  2. empowering LoRa Gateways to act as routers, without the need to employ Network Servers and to tunnel all traffic to or from them;
  3. enabling secure data sharing and wireless Node control; and
  4. maintaining the important power conservation and robustness properties of current LoRaWAN systems.

To achieve these goals without abandoning the benefits of the LoRA PHY (i.e., a robust, energy-efficient long-range communication channel) we developed both a complete redesign of the MAC layer and a data-oriented network layer on top. Our work leverages two key building blocks.

  1. the Deterministic and Synchronous Multi-Channel Extension (DSME) extension to IEEE 802.15.4e, a flexible MAC layer that consists of contention-access and contention-free periods, and,
  2. the Information-Centric Networking (ICN) protocol NDN, which provides secure access to named data in networks.

LoRa and ICN

Prior work showed that ICN provides clear benefits over traditional IP and CoAP or MQTT stacks in the IoT. Our research showed that ICN is also well-suited for LoRa networks because its hop-wise data replication increases robustness and flexibility while reducing retransmission load. This enhances adaptivity and decreases communication overhead, whereas link capacity is scarce with LoRa. Named and authenticated data access enables location-independence since applications can access named data directly, without resorting to lower-layer addresses. Furthermore, built-in caches in ICN facilitate more efficient LoRa networks. Requests that are satisfied by an in-network cache

  1. reduce link utilization, to improve on air time and wireless interference;
  2. facilitate Node sleep; and
  3. reduce long round trips introduced by slow transmissions.

Results

In our paper, we describe

  1. the design of ICN over LoRa, including a suitable DSME configuration and options for mapping ICN messages to DSME;
  2. a complete simulation environment in OMNeT++ that combines ccnSim as an ICN stack, openDSME as a MAC layer, and FLoRa to simulate LoRa-type devices—and a demonstration of our adaptation layers in that system.
  3. Preferred mappings and additional Node requirements for implementing relevant ICN interaction patterns, based on our simulation results.

Code and documentation is available at https://github.com/inetrg/IFIP-Networking-LoRa-ICN-2022, and the whole system is currently being implemented for the RIOT Operating System.

References

Written by dkutscher

May 17th, 2022 at 3:01 pm

Posted in Publications

Tagged with , ,