Mesh networking

What is a mobile ad hoc network (MANET)?

A mobile ad hoc network (MANET) is a network of mobile devices that act as routers for each other over wireless links, with no fixed infrastructure. Devices can move freely, so the network's shape changes all the time and routing has to keep up. The IETF's MANET working group publishes routing protocols for these networks, such as AODV and OLSRv2.

Learning objectives

After reading this article you will be able to:

  • Describe the traits RFC 2501 says set mobile ad hoc networks apart
  • Compare reactive and proactive MANET routing using AODV and OLSRv2
  • Distinguish a MANET from a delay-tolerant network

What makes a network a MANET

A mobile ad hoc network is a set of wireless devices that form a network among themselves, with no cell tower, access point or other fixed infrastructure in charge. Each device, called a node, both sends its own traffic and routes traffic for others, so a message can reach a device out of direct range through a chain of multi-hop relays.

The IETF’s RFC 2501, published in January 1999 as its MANET working group was getting started, describes a MANET as an autonomous system of mobile nodes that are free to move about arbitrarily. Nodes can be in or on airplanes, ships, trucks and cars, or carried by people. At any moment the radios that can hear each other form a random, multihop graph, and that graph changes as nodes move or adjust their transmitters.

A MANET may operate in isolation, or it may have gateways to a fixed network. RFC 2501 expects it to work as a stub in that case, carrying traffic to and from its own nodes but not acting as a transit network for others.

The traits that shape MANET design

RFC 2501 lists four characteristics that set MANETs apart from wired networks:

  1. Dynamic topologies. Nodes move, so the topology can change randomly and rapidly, and links may be one-way as well as two-way.
  2. Bandwidth-constrained, variable links. Realised throughput is often much lower than the radio’s maximum rate, and congestion is typically the norm.
  3. Energy-constrained operation. Some or all nodes may run on batteries, so saving energy may be the most important design goal.
  4. Limited physical security. Eavesdropping, spoofing and denial of service are easier than on cables, though decentralised control removes the single points of failure of centralised designs.

Those traits explain most MANET design choices. Routes must be found and repaired quickly, control traffic must be kept small, and nodes must be able to sleep.

How MANETs route

RFC 2501 describes two broad approaches, and the IETF has published protocols for both.

  • Reactive, or on demand. Routes are found only when needed, which saves energy and bandwidth at the cost of a delay before the first packet can be sent. AODV (RFC 3561) works this way. It is intended for mobile nodes in an ad hoc network, adapts quickly to changing links, and uses destination sequence numbers to keep routes loop free.
  • Proactive. Every node keeps routes to every destination ready in advance, which removes that delay but costs airtime even when nothing is being sent. OLSRv2 (RFC 7181) is a proactive protocol for MANETs. Each router picks multipoint relays among its neighbours, and only those relays forward flooded control traffic, which cuts the number of transmissions.

Both depend on knowing who the neighbours are. NHDP (RFC 6130) handles that, discovering one-hop and symmetric two-hop neighbours. How discovery, route choice and loop prevention fit together for a single message is covered in how a message crosses a mesh network.

The IETF MANET working group today

The MANET working group is still active. Its charter is to standardise IP routing for wireless networks with both static and dynamic topologies, using approaches that are relatively lightweight and suit many kinds of hardware and radio. It also covers hybrid meshes that mix fixed and mobile routers.

The group maintains OLSRv2, NHDP and the generalised MANET packet and message format (RFC 5444). Its charter puts special attention on the Dynamic Link Exchange Protocol (DLEP), which lets a router learn the current state of the radio modem it uses, such as link speed, so it can make better routing decisions as conditions change.

MANETs, meshes and delay-tolerant networks

RFC 2501 says mobile ad hoc networking has also been called mobile packet radio networking and mobile mesh networking. In other words, a MANET is a mesh network whose nodes move, with routing at the IP layer.

Phone meshes built by apps share the same problems: moving devices, changing links, limited batteries and untrusted relays. Some solve them in the application rather than in IP routing, but many of the ideas carry over, including hop limits, neighbour discovery and routes that heal around failures.

MANET routing still looks for a path that exists now. When devices are so sparse that no end-to-end path may exist at once, a different approach is needed. RFC 4838 notes that internet protocols assume an end-to-end path exists for the length of a session, and describes delay-tolerant networking, in which nodes store messages and forward them when a contact appears.

Frequently asked questions

Is a MANET the same as a mesh network?

A MANET is a kind of mesh network in which the nodes move. RFC 2501 notes that the field has also been called mobile mesh networking and mobile, multihop, wireless networking.

Can a MANET connect to the internet?

Yes, if some of its nodes act as gateways. RFC 2501 says a MANET may operate in isolation or have gateways to a fixed network, typically acting as a stub network that carries only its own nodes' traffic.

Sources

Build it with Offline Protocol

The transport and routing page describes how the Offline Protocol mesh SDK forwards messages across multiple hops between participating devices, and what gateways to other networks require.

Read the transport and routing docs