The State of Wireless London

01.01.70

Download text ↓

Wireless Freenetworks

commercial-consume

A Free Network is one which is owned and operated by its users, and allows for the free (libre) transit of data across it. By connecting many smaller freenetworks a larger network can be constructed and the sum of these is often termed The Freenetwork. Wireless networking makes this patchwork of networks possible in an urban environment like London by removing the need for running wires over different properties.

Wireless freenetworking as an idea has strong roots in London. The Consume project was one of the first to propose using the 802.11b equipment that had been originally been designed for home and office internal networking for building local user owned infrastructure.

The project came out of 3 years of practical experience that started in 1997 with a link between two buildings on either side of Clink Street. The link between Backspace and Mediumrare provided access to an expensive internet leased line [10]_ using earlier 802.11 equipment. While an expensive solution at the time, it enabled the creation of a local network serving over 100 people. The wireless equipment had been chosen over a cable run in order to avoid a planning law that meant that only Public Telecoms Operators could string a wire across a street.

In 2000 James Stevens and I wrote a manifesto [11]_ and publicised a proposed expansion of this locally tested idea to build a wider high-speed network, owned by its users. Under the banner, ‘Trip the loop, make your switch, consume the net’ the project was critical of, and opposed to, commercial telecoms providers.

Partly conceived as a way to allow digital art practitioners to experiment with higher speed networks than the offerings available at the time could affordably allow, the project was taken up by a strongly technical crowd with strong free software ties. There were difficult technical problems to be overcome at both software and hardware levels.

At the time the project received a huge amount of media attention from the mainstream media, and was mentioned in media outlets as diverse as The Economist, The Wall Street Journal, Scientific American and Slashdot. The idea was a welcome counter-story to the collapsing dot.com and telecoms bubbles, and perhaps chimed more with non-materialist desires after the excesses of the previous years.

The model was quickly replicated elsewhere, especially in the States, and there is now a burgeoning wireless freenetwork community around the world. While Consume was not the first community wireless group – it had been prefigured by the Bay Area Wireless Group [12]_ and others – it was perhaps the first to grab media attention and gain some kind of wider momentum.

In the last year or so, the Freenetworking Movement has begun to coalesce around new concerns, with recent discourses putting freenetworks and ownership and control of media infrastructure in a freedom of expression context. Control of a network means ultimate control over network traffic. Consume argued that the only way to ensure an autonomous media was to ensure that the network ownership was distributed.

At the Open Cultures conference in Vienna, Eben Moglen, General Legal Counsel of the Free Software Foundation [13]_, addressed the role of free networks in delivering freedom of information. He further says that the three components of free information and hence a free media and freedom of expression are free software, free hardware and free networks.[14]_ He acknowledges public access to radio spectrum and wireless networking as being at the centre of attempts to create these new freenetworks.

With both a growing number of people involved in creating freenetworks, and a number of senior commentators looking at commons-based approaches to spectrum regulation and the politics of network ownership, wireless freenetworking is at the centre of a growing and timely discourse.

Mesh

One of the ideas expressed in the Consume Manifesto [11]_ was that of a meshed network. A meshed network is one with no ‘top’ where each node on the network is connected to a number of neighbours offering many possible routes across it. In contrast to a star topology, which has nodes connected to hubs that then link to backbones, a meshed network is resistant to being controlled as constriction at one node causes data to find another route. In a star topology constriction at one central node controls all transfer to nodes towards the edge of the network. Most commercial broadband networks such as ADSL are star topologies rather than mesh ones.

There are a number of ways of achieving a meshed network, either it can be created automatically using ad hoc mesh routing, or it can be managed by a Network Administrator and stitched together with agreements between Network Neighbours.

The original idea of Consume was to create a metropolitan meshed network that would link users at the edge of the network together into a coherent local infrastructure. This connection would allow collective bargaining for back haul bandwidth, and a free local infrastructure that could support local content and an autonomous media. Finally, this meshed-edge network would provide a challenge to existing telecoms providers by being able to escape from the star topology and its built-in control points. Using an agreement between local network neighbours, the plan was to encourage a systemic de-centralisation and distribution of network ownership and operation.

While there was a huge amount of activity by Freenetworkers, a managed or ad-hoc meshed network is not what emerged in London.

What actually happened

What actually happened was that wireless freenetworks sprung up around activists, and, as the level of technical ability needed to install and operate nodes was initially high, this group of activists were people with a high degree of technical expertise. I’ll term this group Geek Activists.

To see whether a metropolitan meshed network could have emerged from this group, let’s look at its population density. To cover the whole of London in a meshed network would need an overlapping patchwork of wireless networks. To see if this is realistic given the level of interest in wireless freenetworking in London here is a rough look at the density of activity to date.

Reach of an omni directional antenna                    		~ 0.5km
Area covered by a node                                  		~ 0.8 km -2
Assuming one geek activist per node:
Geek activist density for full coverage of an area      		~ 1.25 gakm -2
London area (that enclosed by M25)                      		~ 1250 km2
Number of geek activists for  London coverage				~ 1560
Number of people on the consume mailing list nationally			~ 800
Proportion of UK residents living in London             		~ 0.18
Number of consumers in London                           		~ 145
Number of registered consume nodes in London				~ 142
Current geek activist density                           		~ 0.11

Here we see that while the density of activists needed to cover London is 1.25 the current density is only 0.11. There is a large shortfall of nodes or distribution of aptitude and interest if a city-wide meshed network is to be built.

It should be noted that this calculation is extremely approximate. It does not take into account clustering and we find that activity is often centred in particular areas where there are higher densities of interested artistic and technically minded people. An example of a cluster is in the East End [15]_ where a strong group of networks has been built. The other major uncertainty is in the accuracy of public node databases, as at present there is no direct way of telling how much of presented data is stale.

Access point

What we see as a result of this lack of node density is that the network topology that has mainly been adopted is not the meshed model, which requires an even distribution of nodes, but a star topology. I will call this the Access Point Model.

An access point is simple wireless network device that is not capable of forming a meshed network due to lack of processing power and configurability. The access point is typically connected to an ADSL line and distributes the connection to those nearby. In this model it does not connect with its local network peers to form a local infrastructure, but does enable access to the internet by groups or people in a neighbourhood.

This treats network access as a service like electricity or water. It is a centralised resource that is distributed outwards to ‘consumers’ at the edge of the network. As the gate keeper is the telco providing the ADSL connection, these networks are in effect extensions of the telco provider network rather than an alternative to it.While these networks are often offered for free for public use by their owners, this is often in breach of ADSL providers terms and conditions and as such these networks are susceptible to legal challenges ( as has happened in the US), and price control.

In fact by allowing users to share back haul costs, they really lower the cost of buying ADSL, and provide a lower cost entry point for telecoms providers. This is understood by some ADSL providers and in Germany telcos bundle access points with ADSL, selling 70,000 units in 2003. [16]_

While these networks do not in the long term do much to affect the ownership structure of the network, they have provided a huge wireless cloud that can be accessed by anybody, much of it offered for free by the network owners. In this way the collection of all these networks has created a de facto freenetwork, although not one that is autonomous. Many of the nodes installed are however reconfigurable and, given a significant shift in the density of installations, could be converted to form a contiguous meshed edge network.

Towards a City-Wide Meshed Network

The core problem is one of density. Not enough people are participating in freenetworking for local interconnection to occur. For the situation to change, the number of freenetworkers per square kilometer needs to increase. How can this be achieved?

Broadly, barriers to uptake need to be lowered, in particular the technical barriers. Running an internet network typically requires some specialist knowledge, perhaps an awareness of routing or Unix usage. While attempts to educate people, through clinics such as the Consume Clinics [17]_ or programs such as the University of Openess [18]_ are to be applauded, realistically these will not cause a quick shift in the level of knowledge in the population at large.

Another way round this is to productise and automate, and this is happening both with products from the major suppliers, but also more specialist devices from Locust World [19]_, or 4G systems [20]_, that require little or no configuration and automatically build local meshes. Recently major manufacturers such as Linksys have started shipping Linux based access points (wet54g) [21]_, which may open the way for extremely low-cost devices capable of running a mesh network.

A second barrier is the social barrier. In London in particular, people tend not to know their neighbours with communities often grouped around interest rather than location, and this contributes to forming a social barrier to local connection. Node databases are designed to enable people to locate network peers but this tends to locate only those who already know about freenetworking. To reach a wider audience formal agreements such as the Picopeering Agreement [61]_ are being created. These can provide a ready-made framework for people to negotiate local network connections, and a means of explanation by clearly representing the goals and conditions of participation in the network upfront.

Cost barriers are appreciably falling. WLAN technology is on the same ever-downward cost curve and ever-upward performance curve as computers that is outlined in Moore’s Law [22]_. The final barrier is one of visibility and awareness. There is a perception that wireless freenetworking is a purely technical pursuit, in part due to the early focus on solution of technical issues over the last years. The question of why you would want to build a freenetwork and what you might use it for is largely unasked, at least in the public eye.

A strong focus on what can be done with freenetworks, and an exploration of the spaces opened up creatively and otherwise, would help to drive adoption of the freenetwork, which in turn would create more space for activity.

Given a greater density of participants, a meshed metropolitan network could emerge and give rise to a larger freenetwork, and an autonomous media infrastructure might become possible.