Testbeds' characteristics summary

Commercial model/promoter	research / academic entity
Goals and motivation	research
Administration	centralized initially, but it may end up in a distributed model.
Technologies	wired and wireless
Typical scenarios	urban and rural

In some cases, the initiative to start the network is not from the community, but from a research entity (e.g. a university), with the aim of using it for research purposes [Samanta], [Bernardi].

The administration of these networks may start being centralized in most cases (administered by the academic entity) and may end up in a distributed model in which other local stakeholders assume part of the network administration [Rey].

3. Scenarios where Alternative Networks are deployed

Alternative Network Deployments are present in every part of the world. Even in some high-income countries, these networks have been built as an alternative to commercial ones managed by traditional network operators. This section discusses the scenarios where Alternative Networks have been deployed.

3.1. Digital Divide and Alternative Networks

Although there is no consensus on a precise definition for the term "developing country", it is generally used to refer to nations with a relatively lower standard of living. Developing countries have also been defined as those which are in transition from traditional lifestyles towards the modern lifestyle which began in the Industrial Revolution. When it comes to quantify to which extent a country is a developing country, the Human Development Index has been proposed by the United Nations in order to consider the Gross National Income (GNI), the life expectancy and the education level of the population in a single indicator. Additionally, the Gini Index (World Bank estimate) may be used to measure the inequality, as it estimates the dispersion of the national income (see http://data.worldbank.org/indicator/SI.POV.GINI).

However, at the beginning of the 90's the debates about how to quantify development in a country were shaken by the appearance of Internet and mobile phones, which many authors consider the beginning of the Information Society. With the beginning of this Digital Revolution, defining development based on Industrial Society concepts started to be challenged, and links between digital development and its impact on human development started to flourish. The following dimensions are considered to be meaningful when measuring the digital development state of a country: infrastructures (availability and affordability); ICT (Information and Communications Technology) sector (human capital and technological industry); digital literacy; legal and regulatory framework; and content and services. The lack or less extent of digital development in one or more of these dimensions is what has been referred as Digital Divide. This divide is a new vector of inequality which - as it happened during the Industrial Revolution - generates a lot of progress at the expense of creating a lot economic poverty and exclusion. The Digital Divide is considered to be a consequence of other socio-economic divides, while, at the same time, a reason for their rise.

In this context, the so-called “developing countries”, in order not to be left behind of this incipient digital revolution, motivated the World Summit of the Information Society which aimed at achieving “a people-centred, inclusive and development-oriented Information Society, where everyone can create, access, utilize and share information and knowledge, enabling individuals, communities and peoples to achieve their full potential in promoting their sustainable development and improving their quality of life” [WSIS], and called upon “governments, private sector, civil society and international organizations” to actively engage to accomplish it [WSIS].

Most efforts from governments and international organizations focused initially on improving and extending the existing infrastructure in order not to leave their population behind. As an example, one of the goals of the Digital Agenda for Europe [DAE] is "to increase regular internet usage from 60% to 75% by 2015, and from 41% to 60% among disadvantaged people."

Universal Access and Service plans have taken different forms in different countries over the years, with very uneven success rates, but in most cases inadequate to the scale of the problem. Given its incapacity to solve the problem, some governments included Universal Service and Access obligations to mobile network operators when liberalizing the telecommunications market. In combination with the overwhelming and unexpected uptake of mobile phones by poor people, this has mitigated the low access indicators existing in many developing countries at the beginning of the 90s [Rendon].

Although the contribution made by mobile network operators in decreasing the access gap is undeniable, their model presents some constraints that limit the development outcomes that increased connectivity promises to bring. Prices, tailored for the more affluent part of the population, remain unaffordable to many, who invest large percentages of their disposable income in communications. Additionally, the cost of prepaid packages, the only option available for the informal economies existing throughout developing countries, is high compared with the rate longer-term subscribers pay.

The consolidation of many Alternative Networks (e.g. Community Networks) in high income countries sets a precedent for civil society members from the so-called developing countries to become more active in the search for alternatives to provide themselves with affordable access. Furthermore, Alternative Networks could contribute to other dimensions of the digital development like increased human capital and the creation of contents and services targeting the locality of each network.

3.2. Urban vs. rural areas

The Digital Divide presented in the previous section is not only present between countries, but within them too. This is specially the case for rural inhabitants, which represents approximately 55% of the world's population, from which 78% inhabit in developing countries. Although it is impossible to generalize among them, there exist some common features that have determined the availability of ICT infrastructure in these regions. The disposable income of their dwellers is lower than those inhabiting urban areas, with many surviving on a subsistence economy. Many of them are located in geographies difficult to access and exposed to extreme weather conditions. This has resulted in the almost complete lack of electrical infrastructure. This context, together with their low population density, discourages telecommunications operators to provide similar services to those provided to urban dwellers, since they do not deem them profitable.

The cost of the wireless infrastructure required to set up a network, including powering it (e.g. via solar energy), is within the range of availability if not of individuals at least of entire communities. The social capital existing in these areas can allow for Alternative Network set-ups where a reduced number of nodes may cover communities whose dwellers share the cost of the infrastructure and the gateway and access it via inexpensive wireless devices. Some examples are presented in [Pietrosemoli] and [Bernardi].

In this case, the lack of awareness and confidence of rural communities to embark themselves in such tasks can become major barriers to their deployment. Scarce technical skills in these regions have also been pointed as a challenge for their success, but the proliferation of urban Community Networks, where scarcity of spectrum, scale, and heterogeneity of devices pose tremendous challenges to their stability and the services they aim to provide, has fuelled the creation of robust low-cost low-consumption low-complexity off-the-shelf wireless devices which make much easier the deployment and maintenance of these alternative infrastructures in rural areas.

3.3. Gap between demanded and provided communications services

Beyond the Digital Divide, either international or domestic, there are many situations in which the market fails to provide the information and communications services demanded by the population. When this happens permanently in an area, citizens may be compelled to take a more active part in the design and implementation of ICT solutions, hence promoting Alternative Networks.

3.4. Topology patterns followed by Alternative Networks

Alternative Networks, considered self-managed and self-sustained, follow different topology patterns [Vega]. Generally, these networks grow spontaneously and organically, that is, the network grows without specific planning and deployment strategy and the routing core of the network fits fairly well a power law distribution. Moreover, the network is composed of a high number of heterogeneous devices with the common objective of freely connecting and increasing the network coverage. Although these characteristics increase the entropy (e.g., by increasing the number of routing protocols), they have resulted in an inexpensive solution to effectively increase the network size. One example corresponds to Guifi.net [Vega] with an exponential grow rate in the number of operating nodes during the last decade.

Regularly rural areas in these networks are connected through long-distance links (the so-called community mesh approach) which in turn convey the Internet connection to relevant organisations or institutions. In contrast, in urban areas, users tend to share and require mobile access. Since these areas are also likely to be covered by commercial ISPs, the provision of wireless access by Virtual Operators like [Fon] may constitute a way to extend the user capacity (or gain connection) to the network. Other proposals like Virtual Public Networks [Sathiaseelan_a] can also extend the service.

4. Technologies employed

4.1. Wired

In many (developed or developing) countries it may happen that national service providers may decline to provide connectivity to tiny and isolated villages. So in some cases the villagers have created their own optical fiber networks. It is the case of Lowenstedt in Germany [Lowenstedt], or some parts of Guifi.net [Cerda-Alabern].

4.2. Wireless

The vast majority of the Alternative Network Deployments are based on different wireless technologies [WNDW]. Below we summarize topics to be considered in such deployments. Different considerations about the available options are presented, including physical and Media Access Control (MAC) layers. In addition, the trends (and some recommendations) when using these features in Alternative Networks are summarized.

4.2.1. Antennas

Three kinds of antennas are suitable to be used in these networks: omnidirectional, low-gain directional and high-gain directional antennas.

For local access, omnidirectional antennas are the most useful, since they provide the same coverage in all directions of the plane in which they are located. Above and below this plane, the received signal will diminish, so the maximum benefits are obtained when the client is at approximately the same height as the Access Point.

For indoor clients, omnidirectional antennas are generally fine, because the numerous reflections normally found in indoor environments negate the advantage of using directional antennas.

For outdoor clients, directional antennas can be quite useful to extend coverage to an Access Point fitted with an omnidirectional one.

When building point-to-point links, the highest gain antennas are the best choice, since their narrow beamwidth mitigates interference from other users and can provide the longest links [Flickenger], [Zennaro].

Despite the fact that the free space loss is directly proportional to the square of the frequency, it is normally advisable to use higher frequencies for point-to-point links when there is a clear line of sight, because it is normally easier to get higher gain antennas, the protection against interferences is better and the spectrum saturation is lower. Deploying high gain antennas at both ends will more than compensate for the additional free space loss.

On the contrary, lower frequencies offer advantages when the line of sight may be blocked because they can leverage diffraction to reach the intended receiver.

In the case of mesh networking, where the antenna should connect to several other nodes, it is better to use omnidirectional antennas.

The same type of polarization must be used at both ends of any radio link. For point-to-point links, parabolic antennas exist that may transmit/receive two different signals simultaneously at the same frequency but with orthogonal polarizations, thus permitting to increase the achievable throughput significantly and to improve the protection to multipath and to other transmission impairments.

4.2.2. Physical link length

4.2.2.1. Line-of-Sight

For short distance transmission, there is no strict requirement of line of sight between the transmitter and the receiver, and multipath can guarantee communication despite the existence of obstacles in the direct path.

For longer distances, the first requirement is the existence of an unobstructed line of sight between the transmitter and the receiver. For very long paths, the earth curvature is an obstacle that must be cleared, but the trajectory of the radio beam is not strictly a straight line due to the bending of the rays as a consequence of non- uniformities of the atmosphere. Most of the time this bending will mean that the radio horizon extends further than the optical horizon.

4.2.2.2. Transmitted and Received Power

Once a clear radio-electric line of sight is obtained, it is required that the received power is significantly above the sensitivity of the receiver, by what is known as the "link margin". The greater the link margin, the more reliable the link. For mission critical applications 20 dB margin is suggested, but for non critical ones 10 dB might suffice.

The sensitivity of the receiver decreases with the transmission speed, so more power is needed at greater transmission speeds.

Different options can be considered in order to achieve a given link margin (also called "fade margin"):

a) To increase the output power.The maximum transmitted power is specified by each country's regulation, and for unlicensed frequencies is much lower than for licensed frequencies.

b) To increase the antenna gain. There is no limit in the gain of the receiving antenna, but high gain antennas are bulkier, present more wind resistance and require sturdy mounts to comply with tighter alignment requirements. The transmitter antenna gain is also regulated and can be different for point-to-point as for point-to- multipoint links.

c) To reduce the propagation loss, by using a more favorable frequency or a shorter path.

d) To use a more sensitive receiver. Receiver sensitivity can be improved by using better circuits, but it is ultimately limited by the thermal noise, which is proportional to temperature and bandwidth. One can increase the sensitivity by using a smaller receiving bandwidth, or by settling to lower throughput even in the same receiver bandwidth.

4.2.3. Media Access Control (MAC) Protocols for Wireless Links

Different protocols for Media Access Control, which also include physical layer (PHY) recommendations, are widely used in Alternative Network Deployments. Wireless standards ensure interoperability and usability to those who design, deploy and manage wireless networks.

The standards used in the vast majority of Alternative Networks come from the IEEE Standard Association's IEEE 802 Working Group. Standards developed by other international entities can also be used, as e.g. the European Telecommunications Standards Institute (ETSI).

4.2.3.1. 802.11 (Wi-Fi)

The standard we are most interested in is 802.11 a/b/g/n/ac, as it defines the protocol for Wireless LAN. It is also known as "Wi-Fi". The original release (a/b) was issued in 1999 and allowed for rates up to 54 Mbit/s. The latest release (802.11ac) issued in 2011 reaches up to 866.7 Mbit/s. In 2012, the IEEE issued the 802.11-2012 Standard that consolidates all the previous amendments. The document is freely downloadable from IEEE Standards [IEEE].

4.2.3.1.1. Deployment planning for 802.11 wireless networks

To provide physical connectivity, wireless network devices must operate in the same part of the radio spectrum. This means that 802.11a radios will talk to 802.11a radios at around 5 GHz, and 802.11b/g radios will talk to other 802.11b/g radios at around 2.4 GHz. But an 802.11a device cannot interoperate with an 802.11b/g device, since they use completely different parts of the electromagnetic spectrum. More specifically, wireless interfaces must agree on a common channel. If one 802.11b radio card is set to channel 2 while another is set to channel 11, then the radios cannot communicate with each other.

Each 802.11 device can operate in one of four possible modes:

1. Master mode (also called AP or infrastructure mode) is used to create a service that looks like a traditional Access Point. Wireless interfaces in master mode can only communicate with interfaces that are associated with them in managed mode.

2. Managed mode is sometimes also referred to as client mode. Wireless interfaces in managed mode will join a network created by a master, and will automatically change their channel to match it. Managed mode interfaces do not communicate with each other directly, and only communicate with an associated master.

3. Ad-hoc mode creates a multipoint-to-multipoint network where there is no single master node or AP. In ad-hoc mode, each wireless interface communicates directly with its neighbors. Ad-hoc mode is often also called Mesh Networking.

4. Monitor mode is used by some tools (such as Kismet) to passively listen to all radio traffic on a given channel.

When implementing a point-to-point or point-to-multipoint link, one radio will typically operate in master mode, while the other(s) operate in managed mode. Ad-hoc is more flexible but has a number of performance issues as compared to using the master / managed modes.

4.2.3.1.2. Long Distances in 802.11

The MAC protocol in 802.11 is called CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) and was designed for short distances; the transmitter expects the reception of an acknowledgment for each transmitted unicast packet; if a certain waiting time is exceeded, the packet is retransmitted. This behavior makes necessary the adaptation of several MAC parameters when 802.11 is used in long links [Simo_b]. Even with this adaptation, the distance has a significant negative impact on the performance. For this reason, many vendors implement alternative medium access techniques that are offered alongside the standard CSMA/CA in their outdoor 802.11 products. These alternative proprietary MAC protocols usually employ some type of TDMA (Time Division Multiple Access). Low cost equipment using these techniques can offer high throughput at distances above 100 kilometers.

4.2.3.2. GSM

GSM (Global System for Mobile Communications), from ETSI, has also been used in Alternative Networks as Layer 2 option, as explained in [Mexican].

4.2.3.3. Dynamic Spectrum

Some Alternative Networks make use of TV White Spaces – a set of UHF and VHF television frequencies that can be utilized by secondary users in locations where it is unused by licensed primary users such as television broadcasters. Equipment that makes use of TV White Spaces is required to detect the presence of existing unused TV channels by means of a spectrum database and/or spectrum sensing in order to ensure that no harmful interference is caused to primary users. In order to smartly allocate interference-free channels to the devices, cognitive radios are used which are able to modify their frequency, power and modulation techniques to meet the strict operating conditions required for secondary users.

The use of the term “White Spaces” is often used to describe “TV White Spaces” as the VHF and UHF television frequencies were the first to be exploited on a secondary use basis. There are two dominant standards for TV white space communication: (i) the 802.11af standard [IEEE.802-11AF.2013] – an adaptation of the 802.11 standard for TV white space bands and (ii) the IEEE 802.22 standard [IEEE.802-22.2011] for long-range rural communication.

4.2.3.3.1. 802.11af

802.11af [IEEE.802-11AF.2013] is a modified version of the 802.11 standard operating in TV White Space bands using Cognitive Radios to avoid interference with primary users. The standard is often referred to as White-Fi or "Super Wi-Fi" and was approved in February 2014. 802.11af contains much of the advances of all the 802.11 standards including recent advances in 802.11ac such as up to four bonded channels, four spatial streams and very high rate 256-QAM modulation but with improved in-building penetration and outdoor coverage. The maximum data rate achievable is 426.7 Mbps for countries with 6/7 MHz channels and 568.9 Mbps for countries with 8 MHz channels. Coverage is typically limited to 1km although longer range at lower throughput and using high gain antennas will be possible.

Devices are designated as enabling stations (Access Points) or dependent stations (clients). Enabling stations are authorized to control the operation of a dependent station and securely access a geolocation database. Once the enabling station has received a list of available white space channels it can announce a chosen channel to the dependent stations for them to communicate with the enabling station. 802.11af also makes use of a registered location server – a local database that organizes the geographic location and operating parameters of all enabling stations.

4.2.3.3.2. 802.22

802.22 [IEEE.802-22.2011] is a standard developed specifically for long range rural communications in TV white space frequencies and first approved in July 2011. The standard is similar to the 802.16 (WiMax) [IEEE.802-16.2008] standard with an added cognitive radio ability. The maximum throughput of 802.22 is 22.6 Mbps for a single 8 MHz channel using 64-QAM modulation. The achievable range using the default MAC scheme is 30 km, however 100 km is possible with special scheduling techniques. The MAC of 802.22 is specifically customized for long distances – for example, slots in a frame destined for more distant Consumer Premises Equipment (CPEs) are sent before slots destined for nearby CPEs.

Base stations are required to have a Global Positioning System (GPS) and a connection to the Internet in order to query a geolocation spectrum database. Once the base station receives the allowed TV channels, it communicates a preferred operating white space TV channel with the CPE devices. The standard also includes a co-existence mechanism that uses beacons to make other 802.22 base stations aware of the presence of a base station that is not part of the same network.

5. Upper layers

5.1. Layer 3

5.1.1. IP addressing

Most known Alternative Networks started in or around the year 2000. IPv6 was fully specified by then, but almost all Alternative Networks still use IPv4. A survey [Avonts] indicated that IPv6 rollout presents a challenge to Community Networks.

Most Community Networks use private IPv4 address ranges, as defined by [RFC1918]. The motivation for this was the lower cost and the simplified IP allocation because of the large available address ranges.

5.1.2. Routing protocols

As stated in previous sections, Alternative Networks are composed of possibly different layer 2 devices, resulting in a mesh of nodes. Connection between different nodes is not guaranteed and the link stability can vary strongly over time. To tackle this, some Alternative Networks use mesh network routing protocols while other networks use more traditional routing protocols. Some networks operate multiple routing protocols in parallel. For example, they use a mesh protocol inside different islands and use traditional routing protocols to connect these islands.

5.1.2.1. Traditional routing protocols

The Border Gateway Protocol (BGP), as defined by [RFC4271] is used by a number of Community Networks, because of its well-studied behavior and scalability.

For similar reasons, smaller networks opt to run the Open Shortest Path First (OSPF) protocol, as defined by [RFC2328].

5.1.2.2. Mesh routing protocols

A large number of Alternative Networks use the Optimized Link State Routing Protocol (OLSR) routing protocol as defined in [RFC3626]. The pro-active link state routing protocol is a good match with Alternative Networks because it has good performance in mesh networks where nodes have multiple interfaces.

The Better Approach To Mobile Adhoc Networking (BATMAN) [Abolhasan] protocol was developed by members of the Freifunk community. The protocol handles all routing at layer 2, creating one bridged network.

Parallel to BGP, some networks also run the BatMan-eXperimental (BMX6) protocol [Neumann]. This is an advanced version of the BATMAN protocol which is based on IPv6 and tries to exploit the social structure of Alternative Networks.

5.2. Transport layer

5.2.1. Traffic Management when sharing network resources

When network resources are shared, a special care has to be put on the management of the traffic at upper layers. From a crowdshared perspective, and considering just regular TCP connections during the critical sharing time, the Access Point offering the service is likely to be the bottleneck of the connection. This is the main concern of sharers, having several implications. There should be an adequate Active Queue Management (AQM) mechanism that implements a Lower-than-best-effort (LBE) [RFC6297] policy for the user and protects the sharer. Achieving LBE behavior requires the appropriate tuning of the well known mechanisms such as Explicit Congestion Notification (ECN) [RFC3168], or Random Early Detection (RED) [RFC2309], or other more recent AQM mechanisms such as Controlled Delay (CoDel) and [I-D.ietf-aqm-codel] PIE (Proportional Integral controller Enhanced) [I-D.ietf-aqm-pie] that aid on keeping low latency.

However, other bottlenecks besides client’s access bottleneck may not be controlled by the previously mentioned protocols. Therefore, recently proposed transport protocols like LEDBAT [Ros], [Komnios] with the purpose of transporting scavenger traffic may be a solution. LEDBAT requires the cooperation of both the client and the server to achieve certain target delay, therefore controlling the impact of the user along all the path.

There are applications that manage aspects of the network from the sharer side and from the client side. From sharer's side, there are applications to centralize the management of the APs conforming the network that have been recently proposed by means of SDN [Sathiaseelan_a], [Suresh]. There are also other proposals such as Wi2Me [Lampropulos] that manage the connection to several Community Networks from the client’s side. These applications have shown to improve the client performance compared to a single-Community Network client.

5.2.2. Multi-hop issues

On the other hand, transport protocols inside a multiple hop wireless mesh network are likely to suffer performance degradation for multiple reasons, e.g., hidden terminal problem, unnecessary delays on the TCP ACK clocking that decrease the throughout or route changing [Hanbali]. There are some options for network configuration. The implementation of an easy-to-adopt solution for TCP over mesh networks may be implemented from two different perspectives. One way is to use a TCP-proxy to transparently deal with the different impairments ([RFC3135]). Another way is to adopt end-to-end solutions for monitoring the connection delay so that the receiver adapts the TCP reception window (rwnd) [Castignani_c]. Similarly, the ACK Congestion Control (ACKCC) mechanism [RFC5690] could deal with TCP-ACK clocking impairments due to inappropriate delay on ACK packets. ACKCC compensates in an end-to-end fashion the throughput degradation due to the effect of media contention as well as the unfairness experienced by multiple uplink TCP flows in a congested Wi-Fi access.

5.3. Services provided

This section provides an overview of the services between hosts inside the network. They can be divided into Intranet services, connecting hosts between them, and Internet services, connecting to nodes outside the network.

5.3.1. Intranet services

Intranet services can include, but are not limited to:

- VoIP (e.g. with SIP)

- Remote desktop (e.g. using my home computer and my Internet connection when I am on holidays in a village).

- FTP file sharing (e.g. distribution of Linux software).

- P2P file sharing.

- Public video cameras.

- DNS.

- Online games servers.

- Jabber instant messaging.

- IRC chat.

- Weather stations.

- NTP.

- Network monitoring.

- Videoconferencing / streaming.

- Radio streaming.

5.3.2. Access to the Internet

5.3.2.1. Web browsing proxies

A number of federated proxies may provide web browsing service for the users. Other services (file sharing, skype, etc.) are not usually allowed in many Alternative Networks due to bandwidth limitations.

5.3.2.2. Use of VPNs

Some “micro-ISPs” may use the network as a backhaul for providing Internet access, setting up VPNs from the client to a machine with Internet access.

6. Acknowledgements

This work has been partially funded by the CONFINE European Commission Project (FP7 – 288535). Arjuna Sathiaseelan and Andres Arcia Moret were funded by the EU H2020 RIFE project (Grant Agreement no: 644663). Jose Saldana was funded by the EU H2020 Wi-5 project (Grant Agreement no: 644262).

The editor and the authors of this document wish to thank the following individuals who have participated in the drafting, review, and discussion of this memo:

Paul M. Aoki, Roger Baig, Jaume Barcelo, Steven G. Huter, Rohan Mahy, Rute Sofia, Dirk Trossen.

A special thanks to the GAIA Working Group chairs Mat Ford and Arjuna Sathiaseelan for their support and guidance.

7. Contributing Authors

Leandro Navarro
U. Politecnica Catalunya
Jordi Girona, 1-3, D6
Barcelona  08034
Spain

Phone: +34 934016807
Email: leandro@ac.upc.edu  

Carlos Rey-Moreno
University of the Western Cape
Robert Sobukwe road
Bellville  7535
South Africa

Phone: 0027219592562
Email: crey-moreno@uwc.ac.za  

Ioannis Komnios
Democritus University of Thrace
Department of Electrical and Computer Engineering
Kimmeria University Campus
Xanthi 67100
Greece

Phone: +306945406585
Email: ikomnios@ee.duth.gr  

Steve Song
Village Telco Limited


Halifax
Canada

Phone:
Email: stevesong@nsrc.org  

David Lloyd Johnson
Meraka, CSIR
15 Lower Hope St
Rosebank 7700
South Africa

Phone: +27 (0)21 658 2740
Email: djohnson@csir.co.za  

Javier Simo-Reigadas
Escuela Técnica Superior de Ingeniería de Telecomunicación
Campus de Fuenlabrada
Universidad Rey Juan Carlos
Madrid
Spain

Phone: 91 488 8428 / 7500
Email: javier.simo@urjc.es  

8. IANA Considerations

This memo includes no request to IANA.

9. Security Considerations

No security issues have been identified for this document.

10. References

10.1. Normative References

, ", ", "

[IEEE]	Institute of Electrical and Electronics Engineers, IEEE, IEEE Standards association", 2012.
[IEEE.802-11AF.2013]	Information technology - Telecommunications and information exchange between systems - Local and metropolitan area networks - Specific requirements - Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications - Amendment 5: Television White Spaces (TVWS) Operation", IEEE Standard 802.11af, Oct 2009.
[IEEE.802-16.2008]	Information technology - Telecommunications and information exchange between systems - Broadband wireless metropolitan area networks (MANs) - IEEE Standard for Air Interface for Broadband Wireless Access Systems", IEEE Standard 802.16, Jun 2008.
[IEEE.802-22.2011]	Information technology - Telecommunications and information exchange between systems - Local and metropolitan area networks - Specific requirements - Part 22: Cognitive Wireless RAN Medium Access Control (MAC) and Physical Layer (PHY) specifications: Policies and procedures for operation in the TV Bands", IEEE Standard 802.22, Jul 2011.
[RFC1918]	Rekhter, Y., Moskowitz, R., Karrenberg, D., Groot, G. and E. Lear, "Address Allocation for Private Internets", BCP 5, RFC 1918, February 1996.
[RFC2309]	Braden, B., Clark, D., Crowcroft, J., Davie, B., Deering, S., Estrin, D., Floyd, S., Jacobson, V., Minshall, G., Partridge, C., Peterson, L., Ramakrishnan, K., Shenker, S., Wroclawski, J. and L. Zhang, "Recommendations on Queue Management and Congestion Avoidance in the Internet", RFC 2309, April 1998.
[RFC2328]	Moy, J., "OSPF Version 2", STD 54, RFC 2328, April 1998.
[RFC3135]	Border, J., Kojo, M., Griner, J., Montenegro, G. and Z. Shelby, "Performance Enhancing Proxies Intended to Mitigate Link-Related Degradations", RFC 3135, June 2001.
[RFC3168]	Ramakrishnan, K., Floyd, S. and D. Black, "The Addition of Explicit Congestion Notification (ECN) to IP", RFC 3168, September 2001.
[RFC3626]	Clausen, T. and P. Jacquet, "Optimized Link State Routing Protocol (OLSR)", RFC 3626, October 2003.
[RFC4271]	Rekhter, Y., Li, T. and S. Hares, "A Border Gateway Protocol 4 (BGP-4)", RFC 4271, January 2006.
[RFC5690]	Floyd, S., Arcia, A., Ros, D. and J. Iyengar, "Adding Acknowledgement Congestion Control to TCP", RFC 5690, February 2010.
[RFC6297]	Welzl, M. and D. Ros, "A Survey of Lower-than-Best-Effort Transport Protocols", RFC 6297, June 2011.

10.2. Informative References

[Abolhasan]	Abolhasan, M., Hagelstein, B. and J. Wang, "Real-world performance of current proactive multi-hop mesh protocols", In Communications, 2009. APCC 2009. 15th Asia-Pacific Conference on (pp. 44-47). IEEE. , 2009.
[Airjaldi]	Rural Broadband (RBB) Pvt. Ltd., Airjaldi., "Airjaldi service", Airjaldi web page, www.airjaldi.net , 2015.
[Avonts]	Avonts, J., Braem, B. and C. Blondia, "A Questionnaire based Examination of Community Networks", Proceedings Wireless and Mobile Computing, Networking and Communications (WiMob), 2013 IEEE 8th International Conference on (pp. 8-15) , 2013.
[Bernardi]	Bernardi, B., Buneman, P. and M. Marina, "Tegola tiered mesh network testbed in rural Scotland", Proceedings of the 2008 ACM workshop on Wireless networks and systems for developing regions (WiNS-DR '08). ACM, New York, NY, USA, 9-16 , 2008.
[Braem]	Braem, B., Baig Viñas, R., Kaplan, A., Neumann, A., Vilata i Balaguer, I., Tatum, B., Matson, M., Blondia, C., Barz, C., Rogge, H., Freitag, F., Navarro, L., Bonicioli, J., Papathanasiou, S. and P. Escrich, "A case for research with and on community networks", ACM SIGCOMM Computer Communication Review vol. 43, no. 3, pp. 68-73, 2013.
[Castignani_a]	Castignani, G., Loiseau, L. and N. Montavont, "An Evaluation of IEEE 802.11 Community Networks Deployments", Information Networking (ICOIN), 2011 International Conference on , vol., no., pp.498,503, 26-28 , 2011.
[Castignani_b]	Castignani, G., Monetti, J., Montavont, N., Arcia-Moret, A., Frank, R. and T. Engel, "A Study of Urban IEEE 802.11 Hotspot Networks: Towards a Community Access Network", Wireless Days (WD), 2013 IFIP , pp.1,8, 13-15 , 2013.
[Castignani_c]	Castignani, G., Arcia-Moret, A. and N. Montavont, "A study of the discovery process in 802.11 networks", SIGMOBILE Mob. Comput. Commun. Rev., vol. 15, no. 1, p. 25 , 2011.
[Cerda-Alabern]	Cerda-Alabern, L., "On the topology characterization of Guifi.net", Proceedings Wireless and Mobile Computing, Networking and Communications (WiMob), 2012 IEEE 8th International Conference on (pp. 389-396) , 2012.
[DAE]	European Commission, EC., "A Digital Agenda for Europe", Communication from the Commission of 19 May 2010 to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions – A Digital Agenda for Europe , 2010.
[Everylayer]	former Volo Broadband, Everylayer., "Everylayer", Everylayer web page, http://www.everylayer.com/ , 2015.
[Flickenger]	Flickenger, R., Okay, S., Pietrosemoli, E., Zennaro, M. and C. Fonda, "Very Long Distance Wi-Fi Networks", NSDR 2008, The Second ACM SIGCOMM Workshop on Networked Systems for Developing Regions. USA, 2008 , 2008.
[FNF]	The Free Network Foundation, FNF., "The Free Network Foundation", The Free Network Foundation web page, https://thefnf.org/ , 2014.
[Fon]	Fon Wireless Limited, Fon., "What is Fon", Fon web page, https://corp.fon.com/en , 2014.
[Hanbali]	Hanbali, A., Altman, E. and P. Nain, "A Survey of TCP over Ad Hoc Networks", IEEE Commun. Surv. Tutorials, vol. 7, pp. 22–36 , 2005.
[Heer]	Heer, T., Hummen, R., Viol, N., Wirtz, H., Gotz, S. and K. Wehrle, "Collaborative municipal Wi-Fi networks-challenges and opportunities", Pervasive Computing and Communications Workshops (PERCOM Workshops), 2010 8th IEEE International Conference on (pp. 588-593). IEEE. , 2010.
[I-D.ietf-aqm-codel]	Nichols, K., Jacobson, V., McGregor, A. and J. Jana, "Controlled Delay Active Queue Management", Internet-Draft draft-ietf-aqm-codel-01, April 2015.
[I-D.ietf-aqm-pie]	Pan, R., Natarajan, P., Baker, F. and G. White, "PIE: A Lightweight Control Scheme To Address the Bufferbloat Problem", Internet-Draft draft-ietf-aqm-pie-01, March 2015.
[Komnios]	Komnios, I., Sathiaseelan, A. and J. Crowcroft, "LEDBAT performance in subpacket regimes", IEEE/IFIP WONS, Austria, April 2014 , 2014.
[Lampropulos]	Lampropulos, A., Castignani, G., Blanc, A. and N. Montavont, "Wi2Me: A Mobile Sensing Platform for Wireless Heterogeneous Networks", 32nd International Conference on Distributed Computing Systems Workshops (ICDCS Workshops), 2012, pp. 108–113 , 2012.
[Lowenstedt]	Huggler, J., "Lowenstedt Villagers Built Own Fiber Optic Network", The Telegraph, 03 Jun 2014, available at http://www.telegraph.co.uk/news/worldnews/europe/germany/10871150/German-villagers-set-up-their-own-broadband-network.html , 2014.
[Mexican]	Varma, S., "Lowenstedt Villagers Built Own Fiber Optic Network", The Times of India, 27 Aug 2013, available at http://timesofindia.indiatimes.com/world/rest-of-world/Ignored-by-big-companies-Mexican-village-creates-its-own-mobile-service/articleshow/22094736.cms , 2013.
[Neumann]	Neumann, A., Lopez, E. and L. Navarro, "An evaluation of bmx6 for community wireless networks", In Wireless and Mobile Computing, Networking and Communications (WiMob), 2012 IEEE 8th International Conference on (pp. 651-658). IEEE. , 2012.
[PAWS]	Sathiaseelan, A., Crowcroft, J., Goulden, M., Greiffenhagen, C., Mortier, R., Fairhurst, G. and D. McAuley, "Public Access WiFi Service (PAWS)", Digital Economy All Hands Meeting, Aberdeen , Oct 2012.
[Pietrosemoli]	Pietrosemoli, E., Zennaro, M. and C. Fonda, "Low cost carrier independent telecommunications infrastructure", In proc. 4th Global Information Infrastructure and Networking Symposium, Choroni, Venezuela , 2012.
[Rendon]	Rendon, A., Ludena, P. and A. Martinez Fernandez, "Tecnologías de la Información y las Comunicaciones para zonas rurales Aplicación a la atención de salud en países en desarrollo", CYTED. Programa Iberoamericano de Ciencia y Tecnología para el Desarrollo , 2011.
[Rey]	Rey-Moreno, C., Bebea-Gonzalez, I., Foche-Perez, I., Quispe-Taca, R., Liñán-Benitez, L. and J. Simo-Reigadas, "A telemedicine WiFi network optimized for long distances in the Amazonian jungle of Peru.", Proceedings of the 3rd Extreme Conference on Communication: The Amazon Expedition, ExtremeCom '11 ACM, 2011.
[Ros]	Ros, D. and M. Welzl, "Assessing LEDBAT's Delay Impact", Communications Letters, IEEE , vol.17, no.5, pp.1044,1047, May 2013 , 2013.
[Samanta]	Samanta, V., Knowles, C., Wagmister, J. and D. Estrin, "Metropolitan Wi-Fi Research Network at the Los Angeles State Historic Park", The Journal of Community Informatics, North America, 4 , May 2008.
[Sathiaseelan_a]	Sathiaseelan, A., Rotsos, C., Sriram, C., Trossen, D., Papadimitriou, P. and J. Crowcroft, "Virtual Public Networks", In Software Defined Networks (EWSDN), 2013 Second European Workshop on (pp. 1-6). IEEE. , 2013.
[Sathiaseelan_b]	Sathiaseelan, A. and J. Crowcroft, "LCD-Net: Lowest Cost Denominator Networking", ACM SIGCOMM Computer Communication Review , Apr 2013.
[Sathiaseelan_c]	Sathiaseelan, A., Mortier, R., Goulden, M., Greiffenhagen, C., Radenkovic, M., Crowcroft, J. and D. McAuley, "A Feasibility Study of an In-the-Wild Experimental Public Access WiFi Network", ACM DEV 5, Proceedings of the Fifth ACM Symposium on Computing for Development, San Jose , Dec 2014 pp 33-42, 2014.
[Simo_a]	Simo-Reigadas, J., Morgado, E., Municio, E., Prieto-Egido, I. and A. Martinez-Fernandez, "Assessing IEEE 802.11 and IEEE 802.16 as backhaul technologies for rural 3G femtocells in rural areas of developing countries", EUCNC 2014 , 2014.
[Simo_b]	Simo-Reigadas, J., Martínez-Fernández, A., Ramos-López, J. and J. Seoane-Pascual, "Modeling and Optimizing IEEE 802.11 DCF for Long-Distance Links", IEEE TRANSACTIONS ON MOBILE COMPUTING, 9(6), pp. 881-896 , 2010.
[Suresh]	Suresh, L., Schulz-Zander, J., Merz, R., Feldmann, A. and T. Vazao, "Towards Programmable Enterprise WLANs with ODIN", In Proceedings of the first workshop on Hot topics in software defined networks (HotSDN '12). ACM, New York, NY, USA, 115-120 , 2012.
[Vega]	Vega, D., Cerda-Alabern, L., Navarro, L. and R. Meseguer, "Topology patterns of a community network: Guifi. net.", Proceedings Wireless and Mobile Computing, Networking and Communications (WiMob), 2012 IEEE 8th International Conference on (pp. 612-619) , 2012.
[WiLD]	Patra, R., Nedevschi, S., Surana, S., Sheth, A., Subramanian, L. and E. Brewer, "WiLDNet: Design and Implementation of High Performance WiFi Based Long Distance Networks", NSDI (Vol. 1, No. 1, p. 1) , Apr 2007.
[WNDW]	Wireless Networking in the Developing World/Core Contributors, "Wireless Networking in the Developing World, 3rd Edition", The WNDW Project, available at wndw.net , 2013.
[WSIS]	International Telecommunications Union, ITU, "Declaration of Principles. Building the Information Society: A global challenge in the new millenium", World Summit on the Information Society, 2003, at http://www.itu.int/wsis, accessed 12 January 2004. , Dec 2013.
[Zennaro]	Zennaro, M., Fonda, C., Pietrosemoli, E., Muyepa, A., Okay, S., Flickenger, R. and S. Radicella, "On a long wireless link for rural telemedicine in Malawi", 6th International Conference on Open Access, Lilongwe, Malawi , Nov 2008.

Authors' Addresses