Vous êtes sur la page 1sur 14

Computer Networks 93 (2015) 359–372

Contents lists available at ScienceDirect

Computer Networks
journal homepage: www.elsevier.com/locate/comnet

Software-defined wireless mesh networks for internet
access sharing
Ahmed Abujoda a,∗, David Dietrich a, Panagiotis Papadimitriou a,
Arjuna Sathiaseelan b

Institute of Communications Technology, Leibniz Universität Hannover, Germany
Computer Laboratory, University of Cambridge, UK

a r t i c l e

i n f o

Article history:
Received 15 February 2015
Revised 31 July 2015
Accepted 3 September 2015
Available online 25 September 2015
Wireless mesh network
Internet access

a b s t r a c t
Universal access to Internet is crucial, and as such, there have been several initiatives to enable
wider access to the Internet. Public Access WiFi Service (PAWS) is one such initiative that takes
advantage of the available unused capacity in home broadband connections and allows Lessthan-Best Effort (LBE) access to these resources, as exemplified by Lowest Cost Denominator
Networking (LCDNet). PAWS has been recently deployed in a deprived community in Nottingham, and, as any crowd-shared network, it faces limited coverage, since there is a single point
of Internet access per guest whose availability depends on user sharing policies.
To mitigate this problem and extend the coverage, we use a crowd-shared wireless
mesh network (WMN), at which the home routers are interconnected as a mesh. Such a
WMN provides multiple points of Internet access and can enable resource pooling across all
available paths to the Internet backhaul. In order to coordinate traffic redirections through
the WMN, we implement and deploy a software-defined WMN (SDWMN) control plane in
one of the CONFINE community networks. We further investigate the potential benefits of a
crowd-shared WMN for public Internet access by performing a comparative study between
a WMN and PAWS. Our experimental results show that a crowd-shared WMN can provide
much higher utilization of the shared bandwidth and can accommodate a substantially larger
volume of guest traffic.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
The Internet has evolved into a critical infrastructure
for education, employment, e-governance, remote health
care, digital economy, and social media. However, the Internet today is facing the challenge of a growing digital
divide, i.e., an increasing disparity between those with and
without Internet access [60]. Access problems often stem
from sparsely spread populations living in physically remote
locations, since it is simply not cost-effective for Internet
Service Providers (ISPs) to deploy the required infrastructure

Corresponding author. Tel.: +49 5117622596.
E-mail address: ahmed.abujoda@ikt.uni-hannover.de (A. Abujoda).

1389-1286/© 2015 Elsevier B.V. All rights reserved.

for broadband Internet access in these areas. Coupled with
physical limitations of terrestrial infrastructures (mainly due
to distance) to provide last mile access, remote communities
also incur higher costs for connection between the exchange
and backbone network when using wired technologies, because the distances are longer. Ubiquitous mobile broadband
coverage is currently not feasible, since direct investment in
local infrastructure is uneconomic [55]. Addressing digital
exclusion due to socio-economic barriers is also important.
The United Nations revealed the global disparity in fixed
broadband access, showing that access to fixed broadband
in some countries costs almost 40–100 times their national
average income [34].
The reluctance of network operators (who are economically motivated) to provide wired and cellular infrastructures

we present our evaluation results and discuss the benefits of a crowd-shared WMN for Internet access sharing. during the PAWS trial deployment. self-organized. a crowd-shared WMN provides extended coverage via multiple points of access for each guest. based on the Lowest Cost Denominator Networking (LCDNet) paradigm [58]. Software defined crowd-shared wireless mesh networks The underlying problem with PAWS or any crowd-shared network is that they serve as single point of Internet access to guests within the coverage of the wireless router and hence. We particularly consider crowd-shared WMNs in residential areas. such as economic constraints placed on home users in underprivileged areas where they are enforced to conserve energy by turning off the routers at nights. which provide more localized communication services. we investigate the potential benefits of extending PAWS or any crowd-shared wireless network to a wireless mesh network (WMN) by interconnecting wireless home routers. since all information can be conveyed to a centralized controller facilitating the coordination of traffic redirections.24]. • We develop an algorithm for the assignment of gateways to flows that require redirection. decisions for traffic redirections should be also based on user sharing policies. when these are disclosed in advance. PAWS adopts an approach of community-wide participation. taking into account the flow demands. Furthermore. or other users either for free or as part of a service offering by the ISP (e. traffic redirection requires the assignment of guest flows to gateways and the selection of paths (through the WMN) that provide sufficient capacity and low delay.360 A. [6. PAWS uses an access point connected to the home router for Internet access sharing). PAWS has been facing ongoing deployment challenges. it was observed that home users did not share their broadband connection over periods at which either the whole bandwidth or all the ports of the home router were needed (i. This paper extends our previous study on softwaredefined WMNs. • We quantify the benefits of a crowd-shared WMN for Internet access sharing using a deployment of our SDN control plane in Athens Wireless Metropolitan Network (AWMN) [4]. Essentially. In this respect. 2. The remainder of the paper is organized as follows. we provide an overview of the software-defined WMN (SDWMN) control plane. Section 6 discusses related work. one access point per guest). These community wireless mesh networks have self-sustainable business models. More precisely. Section 3 elaborates on techniques for guest traffic redirection.. • We generate a model for user sharing patterns based on the router on/off periods captured from the PAWS deployment in Nottingham. 1). / Computer Networks 93 (2015) 359–372 to rural/remote areas have led to several initiatives to build large-scale. because sharers needed all the bandwidth of their broadband connection or due to other reasons. taking advantage of the dense deployment of wireless home routers. the residual bandwidth in the access links and the WMN. PAWS is a crowd-shared network with a single point of access per guest. PAWS is essentially a crowd-shared access network (similar to FON [5]) for under-privileged users in urban and rural communities.g. In Section 4. In Section 5. PAWS was deployed with 20 custommade PAWS routers placed in a deprived community in Nottingham and was also trialed out in rural Wales. and as such. at which the efficiency of a crowd-shared WMN was assessed using simulations [18]. we discuss the implementation of the SDWMN and its deployment in AWMN. we deem a centralized control plane as a more suitable approach to WMN management for Internet access sharing... where home broadband subscribers are enabled to donate controlled but free use of their high-speed broadband Internet to fellow citizens.53]). the periods at which user share their Internet connection). PAWS routers were not available for certain periods. due to the reduced cost of using the unlicensed spectrum [51].e.3. As such. Along these lines.e. Given the amount of information that has to be collected before flow assignments can be made. one of the community networks of the CONFINE project [2. https:// community-lab. These observed user behaviors entail significant challenges for the successful adoption of PAWS. Extending PAWS to a crowd-shared WMN departs from the norm: multiple users from different ISPs form part . Based on our experience from the trial PAWS deployment. To mitigate this problem.e. in which home broadband owners share a portion of their home broadband with friends. The main challenge in the management of such a WMN lies in the coordination of guest traffic redirections. Public Access WiFi Service (PAWS) [53] is a communityled crowd-shared WiFi service that uses a set of techniques that make use of the available unused capacity in home broadband networks and allowing Less-than-Best Effort (LBE) access to these resources. such as limited coverage. our contributions are the following: • We present the design and implementation of a SDN control plane for the coordination of guest traffic redirections through the WMN. stemming from user sharing patterns. Such a network would allow home network users to share part of their own broadband connection with the public for free while also connected to each other as a WMN providing extended coverage (Fig.. we perform a comparative study between a WMN and a crowd-shared network with limited coverage (i. as well as Internet backhaul support via peering agreements with traditional network operators who see such networks as a way to extend their reach at a lower cost.. There are also communityled wireless initiatives such as crowd-shared wireless networks. i. they have no provision to extend the coverage when no bandwidth is being shared. Finally. as well as. the advance knowledge of sharing policies. A potential solution to this problem is to extend the PAWS network as a crowd-shared WMN. In particular. In Section 2.e.net/ In this respect. and decentralized community wireless networks that use WiFi mesh technology (including long distance). Community Lab. Section 7 highlights our conclusions. Internet access sharing is highly dependent on user sharing policies (i. we leverage on software-defined networking (SDN) principles for the control plane design. Abujoda et al. such that the shared bandwidth is efficiently utilized. neighbors.

1. we have deployed a SDWMN control plane in AWMN [4]. It is executed by a SDN controller which has knowledge of the WMN topology and utilization. This raises important questions regarding the operation. policy (e. The algorithm aims at maximizing Internet shared bandwidth utilization by accommodating as large as possible number of flows. the management and operation of a crowd-shared WMN can be outsourced to a third party.A. Traffic redirection Assigning flows to network paths with constrained capacity can be formulated as a multi-commodity flow problem which is known to be NP-hard [33].2). A detailed description of the implementation of the SDWMN control plane and the gateway is given in Section 4. i. The traffic redirection module selects the gateway for Internet access (using the algorithm presented in Section 3) and subsequently creates a tunnel between the home router where the guest is connected and the router assigned as the gateway (home routers are deployed in the AWMN RDs). deployed and managed through an evolutionary SDN control abstraction [54]. 2.. The control plane functionality mainly consists of traffic redirection and monitoring. since the wireless nodes in AWMN are not under our control (as opposed to the RDs). A tunnel is dynamically set up when traffic has to be redirected over that path.. Traffic redirection algorithm We represent the WMN as a weighted undirected graph G = (N. the Internet access link utilizations. where N is the set of nodes and L is the set of links . based on user sharing policies. Freifunk [6]). Hereby. Although originally intended for crowd-shared wireless networks such as PAWS. and the user sharing policies. AWMN consists of more than 1000 wireless nodes and 30 research devices (RDs) that host isolated containers (i. we present a heuristic algorithm for the assignment of the gateway and the path over which the traffic will be redirected through the WMN. we rely on SDN to create the notion of Virtual Public Networks (VPuN).e. while most wireless community WMNs today are operated by a single organization. configuration. SNMP) Research Device Research Device Research Device Community Network Fig. according to the needs of his experiment [24]. Fig. Abujoda et al. 2 illustrates an overview of the SDWMN control plane. and it is torn down when it is no longer needed. The main goal of the control plane is to improve the utilization of shared bandwidth by redirecting guest traffic to one of the home routers through the WMN. crowd-shared home networks created. our approach provides more opportunities for non-governmental organizations and local governments (driven by social goals rather than economic) to become virtual network operators. using the sharing policy expression language presented in [54]). / Computer Networks 93 (2015) 359–372 361 Virtual Network Operator SDN Control Plane Sharer Northbound API Traffic Redirection Monitoring Southbound API (OpenFlow. Crowd-shared WMN for public Internet access. SDN can facilitate the management and operation of wireless networks at large scale. The monitoring module collects statistics about the bandwidth utilization of the WMN and the access links. We first describe the traffic redirection algorithm (Section 3. VPuN can be also used for crowd-shared WMNs. In particular.. 3. Enabling a third party to federate such wireless home networks would reduce the operating expenditures for network operators as well as enable new economic models for revenue generation from currently underutilized infrastructures. As such. of the WMN to provide free Internet connectivity. Software-defined WMN control plane overview. and management of crowd-shared WMNs.1) and then investigate the potential benefits of user sharing policies (Section 3.1. Leveraging on SDN’s centralized control and network-wide visibility. In [54].g. allowing him to express a sharing Research Device Research Device Fig. 3.. Traffic is redirected through a tunnel. As part of the Community-Lab [3]. Compared to existing mesh networks for Internet access sharing (e. enabling resource pooling across multiple home broadband connections based on the prevailing network conditions and usage sharing patterns. L).g. the control plane exposes an interface to each home user. so-called slivers) which can be allocated and controlled by a user. Based on the notion of VPuN. we describe a holistic approach of coupling both social and economic incentives in designing future networks allowing the extension of the stakeholder value chain to include more than the two traditional parties (consumer and Internet Service Provider).e.

3 illustrates the number of traffic redirections for different network loads with and without the knowledge of sharing policies. r). Abujoda et al. we model the availability of the home routers using an on-off Markov chain. guest traffic may have to traverse multiple hops in the WMN. performance degradation. Otherwise. Inputs: G = (N. We run a trace-driven simulation of a crowd-shared WMN to investigate the impact of sharing policies on traffic redirections. adjusting the threshold to 1 restricts the search space to the next-hop routers with sufficient shared bandwidth. reordering [56. This can lead to latency inflation. C(ni )). We use D to represent the set of flow demands that have arrived in the network. We parameterize the exponential distributions using datasets from the PAWS deployment in UK. 3. Flow redirections. Subsequently.e. Initially.2.. (1)). Let Pij denote the set of paths in the network G. Algorithm 1 assigns a gateway and the shortest path to each flow demand from the set D. So far. The algorithm is executed whenever there is insufficient shared bandwidth in the local home network. The available bandwidth C(p) associated to a path p ∈ Pij is given by the minimum residual bandwidth of the links along the path: C ( p) = min C (li j ) li j ∈p (1) We further represent a flow demand for a guest with the tuple d = (nu . excluding the previously selected home router. 3. the traffic . the algorithm performs another iteration for the selection of the gateway. the algorithm selects the home router with the highest access link bandwidth (i. D SORT(D) for each d ∈ D do search ← true M ← {N \ nu } // M : candidate set of routers while (search AND M = ∅) g ← argmaxi∈MC (ni ) for each p ∈ Pug if C ( p) < r then Pug ← Pug \ p end if end for if Pug = ∅ then M ← M \ ng else // gateway assignment gateway ← ng path ← SP(Pug ) search ← false end if end while end for // path assignment between nodes of the set N. Each link lij ∈ L between two nodes ni and nj is associated with the available bandwidth C(lij ). / Computer Networks 93 (2015) 359–372 Algorithm 1 Gateway and path assignment. Sharing policies can be specified using a sharing policy expression language (SPEL). between the local home router (nu ) and the assigned gateway (ng ) that satisfy the flow demand (Eq.57] and in general. Fig. This information is conveyed to the SDN control plane and is used as an additional input for gateway assignment.e. User sharing policies Algorithm 1 assigns flows to gateways taking into account the amount of shared bandwidth and optionally the path hop-count. In particular. where nu ∈ N is the node where the guest device has been attached and r denotes the flow rate. we have not exploited any advance knowledge of user sharing policies for traffic redirection..g. To mitigate this. L). This algorithm carries out the gateway assignment based on the available shared bandwidth of the home routers in the crowd-shared network. SPEL essentially provides the necessary means to the sharers to specify the amount of shared bandwidth as well as the period of sharing. between the pair of nodes ni and nj . When sharing policies are known. forwarding traffic to a router which will shortly become unavailable for Internet access sharing). On and off times are exponentially distributed with mean values μon = 106 min and μo f f = 555 min. Each node ni is associated with an Internet access link whose available shared bandwidth is denoted by C(ni ). Fig. the algorithm identifies the set of paths. This eventually computes the path for the traffic redirection through the WMN. the flow demands are sorted in decreasing order. Pug . Although this can pool resources from all home networks where sharing is permitted achieving efficient utilization of the shared bandwidth. In case there is no such path.362 A. For example.. For each flow demand. the shortest among these paths is being identified based on the number of hops (SP function in the algorithm). we further use a variant of this algorithm by introducing a threshold in the number of hops between the local home router and the assigned gateway. the XML-based schema presented in our previous work [54]. e. delay variation. Such information can prevent wasteful traffic redirections (i.

we use sharing policies in the assignment of guest flows to gateways. (iv) monitoring table. using sharing policies leads to fewer traffic redirections. / Computer Networks 93 (2015) 359–372 363 path 1 Src Sink OVS path 2 delay node (tc) Fig. Implementation 50 We provide a detailed description of the implementation of the SDWMN control plane and gateway in Sections 4. UDP traffic is redirected from path 1 to path 2 at the intermediate node. SDWMN controller Fig. 5. 6): • Gateway registration: Each new gateway joining the crowd-shared network is registered with the creation of a new gateway object in the controller repository. which is a 64 bit unique identifier for each OpenFlow switch instance. for each flow redirection the router sends a OFPT_PACKET_IN message to the controller. in our experimental setup.1. as the PAWS router logs indicate. lower throughput for TCP flows. Since OpenFlow supports only the configuration of switch flow tables. we use XML-RPC interface to install and configure tunnels between the gateways. respectively. UDP port numbers (we use UDP-in-IP tunnelling) and OpenFlow switch port numbers of the installed tunnels. (ii) IP address of the WMN port. We perform additional tests to investigate potential implications. Fig. set up in path 2. with a large number of guest flows per home router.. can lead to congestion control invocations. sends an OFPT_FLOW_MOD message to the router. SDWMN controller.e. TOPP [44]. as such. The redirection is carried out by inserting flow entries to an OpenvSwitch (OVS) datapath that has been set up in the intermediate node. Packet reordering. We further employ a delay node. However. Taking the respective acknowledgment messages into account. First. 6. the knowledge of sharing policies does not yield significant gains. due to traffic redirections. For higher loads. 150 125 100 75 4. which contains the names and OpenFlow switch ports numbers (i. the Internet access links are the bottleneck (since they have been configured at lower capacity than the WMN). traffic redirections generate additional communication overhead between the SDWMN controller and the routers. Considering the identified implications of traffic redirections. in turn. The registration process is triggered by POX when the gateway establishes a new control channel with the controller. due to BW fluctuations and the low degree of accuracy of available BW measurement tools (e. Our SDWMN controller consists of the following modules (Fig. hence. In this setup. is directed to the router with the longest period of availability (among the routers with sufficient shared bandwidth). Specifically. Each gateway object holds information about the gateway’s: (i) datapath ID. DSL port. 5 shows that the number of outof-order packets increases in proportion to the delay difference between the two paths. the WMN port and the guests’ wireless network port). Besides packet reordering. which contains the gateway shared bandwidth utilization and WMN path quality. we use the experimental setup depicted in Fig. which. and IGI/PTR [37]) that was also confirmed by our own tests. and. This will result in the out-of-order delivery of a small number of packets. especially for low and moderate network loads.2. which contains the IP addresses. 3. Our path delay measurements in AWMN indicate that a redirection through the WMN increases the RTT with up to 50 ms. which. 25 0 0 100 200 300 400 delay(old_path) − delay(new_path) (ms) 500 4.. and.A.1 and 4. (iii) ports table. 4. number of packets out−of−order 225 200 175 1 Mbps 2 Mbps 5 Mbps Fig. • Monitoring: This module collects and processes the measurements of shared bandwidth utilization and the WMN paths delay and hop counts for each gateway. Experimental setup for packet reordering evaluation. and as such. . Bandwidth (BW) estimation in a WMN is very challenging. to adjust the delay along this path. such as packet reordering and higher communication overhead. the communication overhead per flow redirection is 490 bytes. Abujoda et al. and (v) tunnel table. According to Fig. 4 to quantify the level of packet reordering. the number of routers with sufficient shared bandwidth diminishes.g. This overhead can increase substantially We implement our controller in Python using POX which provides a framework to implement SDN controllers with OpenFlow protocol interface [43]. in turn. pathload [39].

the DSL port). This message carries the flow matching fields [43] (e.. Subsequently. incoming traffic (green line Fig.364 A. (For interpretation of the references to color in this figure legend. Using the OpenFlow protocol. packet encapsulation modules are installed in the gateway. 4. Tunnels are created by encapsulating guest traffic in UDP packets (i. • Traffic redirection: Flows are redirected using the following modules: • Gateway selection: The gateway selection module is invoked when a gateway sends a OFPT_PACKET_IN message to the controller to announce the arrival of a new flow. 7. the home gateway replies with the switch port number of the new encapsulation module. SDWMN gateway Our SDN gateway exposes to the SDWMN controller an OpenFlow and XML-RPC interface to install tunnels and redirect flows (see Fig. port source/destination) as well as the OpenFlow switch output port number. To calculate the shared bandwidth utilization. This consists in sending the IP address of the assigned gateway to the flow home gateway as well as the source and destination port numbers of the UDP header used for encapsulation. / Computer Networks 93 (2015) 359–372 Fig. All measurements are stored in the monitoring table of the respective gateway object. the flow table configuration module installs the respective forwarding entries in the home and the assigned gateway such that the guest flow is redirected accordingly.. This consists in sending a OFPT_FLOW_MOD message to the home and assigned gateways. Based on the measurements (shared BW utilization and path length) provided by the monitoring module and using the algorithm described in Section 3. • Flow table configuration: This module installs the corresponding flow entries in the home gateway and the assigned gateway using OpenFlow. • Tunnel installation: Using the XML-RPC interface. the gateway selection module assigns a gateway to the new flow. the BW measurement module pulls the network port counters of each home gateway to acquire the accumulated number of bytes received/sent in the network port (e. the reader is referred to the web version of this article. 7) is processed by the decapsulation module.e. The destination IP address of the tunnel header is set to the IP address of the assigned gateway.. a tunnel is being set up (see tunnel installation for further details) between the home gateway (where the flow arrives) and the assigned gateway (where the flow is redirected). the path measurement module uses XMP-RPC to pull the RTT and hop-count measurements of each WMN path to/from each gateway.) the aforementioned WMN bandwidth estimation issues do not affect the accuracy of our measurements.g.2. The gateway replies with OFPT_STATS_REPLY message which carries the latest reads of the gateway rx_bytes and tx_bytes counters. Next. the BW measurement module applies exponential moving average to the counters changes over time. 7). In addition. In particular.g. 7). This information is stored in the tunnel table of the home gateway object to be used later by the flow table configuration module. IP-in-UDP) using the encapsulation module (red line Fig. For each new tunnel a new encapsulation module is created. SDWMN gateway. the controller sends an OpenFlow OFPT_STATS_REQUEST message with OFPST_PORT type and the port number to the gateway. Abujoda et al. Upon successful installation of the tunneling module. source/destination IP. On the other hand. which strips the .

1) and further use simulations with larger WMN topologies (Section 5. 5. we run experiments in AWMN (Section 5. To steer data traffic between the physical ports and the modules (encapsulation/decapsulation). Abujoda et al. not assigned Internet access). although there is high demand for Internet access by guests attached to the various home networks. Upon the expiration of the timer. we restricted the duration of our experiments to 250 s. defined as: ASR(T ) =   T finished t st finished T + tT strejected t st  f inished  .6 s and μo f f = 55. Fig. 10) with an arrival rate of 60 flows per minute. Fig. we scale down the mean values to μon = 10.A. Using MGEN [1]. The instantiation of encapsulation/decapsulation modules is performed by the gateway controller in the following steps: (i) creation of TAP interfaces. The module sends 10 probing packets every 30 s to avoid flooding the network. 8 illustrates the WMN topology measured from a single RD. After downscaling the router on/off periods. we use the model described in Section 3. st are rejected (i. / Computer Networks 93 (2015) 359–372 365 outer header before the traffic is delivered to the output port. Experimental results For our experiments. In contrast. The flows arrive to the network according to a Poisson process. a WMN allows to capitalize the unused capacity and accommodate a larger volume of guest traffic. For router availability pattern. We implemented a tool to analyze and construct the topology out of traceroute measurements. we perform a comparative study of crowdshared WMN against PAWS (or any other crowd-shared network with a single point of access). across 10 runs. according to Fig. 9) and the serving rate (Fig. assigned Internet access) and successfully served without disruption due to router unavailabilre jected denotes the total sizes of the flows at time t which ity. using experimentation we quantify the control communication overhead in terms of BW consumption and flow setup delay. Due to occasional reliability issues in the RDs..5 s. there are several transitions between the router states during each experiment.2.. The path measurement module uses ping and traceroute to collect RTT and hop counts of the WMN paths. The time required to generate the signal is negligible (a few milliseconds).) controller to announce the beginning of an on or off period. we further implement a timer to emulate the on and the off period of the router. Furthermore. A list of gateways IP addresses is passed to this module to obtain the measurements. we investigate the scalability of the crowdshared WMN in terms of BW utilization and serving rate with different network sizes using simulations. We initially measure the shared bandwidth utilization (Fig. 8. (ii) appending TAP interfaces names. the gateway sends a signal to the Fig.1. We further use simulations to study the effect of traffic redirection on latency by measuring the shared BW utilization when applying a threshold to the redirection path length (see Section 3). we use 12 research devices (RDs) to deploy 11 home routers (or gateways) and one SDWMN controller. the reader is referred to the web version of this article. Our main evaluation metrics include: (i) the shared bandwidth utilization across all home routers. This switching module also forwards messages from the SDN controller to the gateway controller module through the XML-RPC server. 9 guest traffic . To study the scalability of our control plane. we use Linux TAP devices.e. respectively. The WMN topology measured using traceroute from a research device (blue circle) toward the other research devices (red circles).2). Both encapsulation and decapsulation modules are implemented using Click [41]. This timer can be also used by the sharers for sharing policy configurations in their home routers. More precisely. we rely on OpenvSwitch [49]. To connect modules to OpenvSwitch.e. we initially measure the utilization level of the shared BW and the accumulated serving rate across time and for different flow arrival rates. (2) denotes the total sizes of the flows at time t where st which are accepted (i. To this end. (For interpretation of the references to color in this figure legend. 5. Using experimentation and simulations. the gateway controller collects and transfers the measurements generated by the path measurement module to the SDN controller. Since running tests for more than 555 min (the mean value of the off period) is not feasible. MAC and IP address of the assigned gateway to Click configuration templates stored on the repository (we use templates to speed up modules instantiation) and (iii) installation of the Click configuration. Finally. Evaluation In this section. and (ii) the accumulated serving rate ASR(T) at time T. we generate guest flows with a rate and lifetime sampled out of a uniform and an exponential distribution. In each gateway. Variations in the measurements across the different runs were insignificant. 9 illustrates a low utilization of the shared bandwidth without a WMN during the whole period.

On the other hand.e.366 A. Abujoda et al. the boxplots in Figs.. as the shared bandwidth utilization and the 1 All boxplots show measurements across the time period of the experiment. 9. For different runs we calculate the average. ranging from 20 to 80 flows per minute. due to the presence of a single point of Internet access for each guest. crowd-shared WMNs can accommodate substantially larger volume of guest traffic. Shared bandwidth utilization for diverse flow arrival rates across 250 s experimentation time. We also define the control communication overhead as the bandwidth consumed to setup flows and monitor the shared bandwidth utilization. We define the setup time as the interval between the flow’s first packet arrival at the gateway and its transmission to the next hop. / Computer Networks 93 (2015) 359–372 Fig. as depicted in Fig. Accumulated serving rate for diverse arrival rates across 250 s experimentation time. Shared bandwidth utilization. the flow arrival rates of 60 flows per minute and beyond lead to request rejections. redirection through the WMN results in the full utilization of the shared bandwidth. the variation is insignificant. each point on the graph represents a different time point. Essentially. 12. Fig.1 These results corroborate the efficiency of the WMN for various traffic loads. Fig. Furthermore. serving rate always remain very high. by effectively pooling resources across all home networks. i. 10. 11. 12. Figs. we measure the flow setup time and the average control communication overhead across a range of flow arrival rates. 10. This is essentially the time incurred for gateway selection and flow entry installation. We further measure the shared bandwidth utilization and serving rate with a wide range of guest traffic demands. We run . This stems from the high utilization of the shared bandwidth. To evaluate the scalability of our control plane. In Fig. Accumulated serving rate. our results show the significant benefit that a WMN can bring into crowd-shared networks. In this respect. 11 and 12 illustrate the shared bandwidth utilization and serving rate with diverse flow arrival rates. Fig. due to the insufficient access link bandwidth. 11 and 12 show poor bandwidth utilization and serving rate without a WMN.

Simulation WMN topology. We use the TFA wireless mesh topology [7] which consists of 21 nodes (Fig. Each generated flow has a rate and lifetime sampled out of a uniform and exponential distribution. The guest flows arrive to the network according to a Poisson process. while the control communication overhead increases linearly and is in the range of tenths of Kbps.9 0. that is exactly the network environment we consider for the proposed crowd-shared WMN.e. Guest flows are granted Internet access either through local routers or by redirection to remote routers based on the algorithm presented in Section 3. 1 Fig. since the bottleneck is the Internet access links.3 0. For each scenario. Shared bandwidth utilization.6 0. / Computer Networks 93 (2015) 359–372 367 Fig.5 0. 13.. 15). Each node in this topology represents a home router with shared Internet access bandwidth. in the periods that each router is active). TFA is an operational network deployed in a densely populated residential area. 16. Simulation results To run tests at larger scale. and their arrival and departure time.4 0. 54 and 10 Mbps. Each simulation run comprises 60 h at a time discretization of 20 min.1 0 0 5 10 15 20 25 30 35 40 45 50 55 60 time (hours) Fig. Figs. Setup time per flow. 15.2 with WMN without WMN 0. The simulation results are in accordance with the results from the community network. whereas each edge is a wireless mesh link. Note that our wireless link model does not consider the impact of interference and distance on the offered link bandwidth.7 0. We simulate the router availability using our model with the mean values obtained from the PAWS datasets. We assume a homogeneous setting where each router has 16 Mbps ADSL downlink and set the shared bandwidth per router to 8 Mbps (i. we developed a simulator (in Python) that models the flow-level behavior of the guest traffic in a crowd-shared WMN. the crowd-shared WMN yields significantly higher utilization of the shared bandwidth and accommodates more traffic in comparison to the crowd-shared network without WMN. the performance of our controller scales with increasing control loads. Fig.A. Control communication overhead. . 13 and 14 show that per flow setup time does not change significantly. Abujoda et al. We run our simulation with mesh link capacity of 200. According to Figs. The flow parameters that we use in our simulations are the average flow rate. respectively. 14.8 0. we perform 20 simulation runs. Flows which cannot be accommodated are rejected. our experiments on a single gateway (the results can be easily extrapolated for more gateways) and generate 100 different flows for each arrival rate. This does not have any impact on the simulation results. Guest flows arrive randomly at different home routers. 5. 16 and 17. shared bandwidth utilization 0.2. According to these results.

flow arrival rate across 60 h simulation time. software-defined wireless networks. 23) with larger WMN sizes. This is further confirmed by the increase in WMN utilization (Fig.4 0. the serving rate reaches a steady state. Related work In the following. respectively. 20.2 with WMN without WMN 0. We further run simulations using synthetic WMN topologies of diverse size and structure generated using NPART [45].7 0.1 with WMN without WMN 0 TFA (21) 50 100 150 200 topologies with different size (# of nodes) Fig. shared bandwidth utilization shared bandwidth utilization 80 Fig.3 max number of hops=1 max number of hops=2 max number of hops=3 no limit 0.3 0. based on real-world WMN deployments (Berlin Freifunk[6]). redirected guest traffic traverses up to 4 hops in the WMN.7 0. We further use a variant of our gateway assignment algorithm (briefly discussed in Section 3).7 0. due to either insufficient bandwidth (with WMN) or changes in sharing policies (without WMN).6 0.4 0. Using NPART and the Berlin WMN model. As shown in Figs. Shared bandwidth utilization for a wide range of guest traffic demands is depicted in Fig. 20 and 21. 18. 0 50 flows arrival rate (flows/min) 0.368 A.8 0. 19 illustrates the shared bandwidth utilization for diverse hop-count threshold values ranging from 1 to 3.4 0. and without any limit in the hop count. . we discuss related work on WMN routing. in both cases.1 0 0 0 5 10 15 20 25 30 35 40 45 50 55 20 60 time (hours) 1 1 0. 150 and 200 nodes. In essence. 17. flow arrival rate. This may become more critical in large WMNs.4 0. in particular. the shared bandwidth utilization and serving rate scale linearly with the increase of the WMN size. The serving rate. 22) and the number of re-assigned flows (Fig.5 0. Fig.8 0. 100.2 0.9 0. Shared bandwidth utilization for diverse hop-count threshold values vs. Shared bandwidth utilization vs. 6.5 0. especially for a single hop.1 10 20 30 40 50 60 70 80 90 100 flows arrrival rate (flows/min) Fig. Accumulated serving rate.6 0. 19. In our simulations.3 0.8 0. and Internet access sharing. where long paths can inflate latency.9 0.6 0.7 0. Eventually.5 0.9 0. Shared bandwidth utilization with different topologies for 60 h simulation time.3 0. For all topologies. 18. which introduces a threshold in the number of hops between the home router and the assigned gateway. there is a trade-off between coverage extension (and thus effective bandwidth utilization) and latency. NPART is a tool for generating realistic WMN topologies.8 0.9 shared banwidth utilization accumulated serving rate 0. Fig. We set the wireless link capacity and shared bandwidth to 54 Mbps and 8 Mbps. we generate flows with arrival rate of 120 flows per minute. we randomly generate 4 additional WMN topologies with 50. which allows us to saturate the WMN capacity with all topologies. since Internet access is permitted only through one of the next-hop home routers.2 0.5 0. Restricting the number of hops has a noticeable impact on bandwidth utilization.2 0. drops as guest flows are not granted Internet access.6 0. Abujoda et al.1 0. / Computer Networks 93 (2015) 359–372 1 1 with WMN without WMN 0.

B.25].8 0. WMN routing protocols (e. 21..g. Deutsche Telekom (DT) [16] and British Telecom (BT) [8] are among the network operators that provide free Internet access to their subscribers via a large number .7 0. Internet access sharing. B. Accumulated serving rate with different topologies.3 0. while we use our algorithm for gateway selection whenever traffic redirection is needed. which propose solutions for mobility management using OpenFlow. low control overhead. DSDV [48]. SrcRR computes routes taking into account the effect of interference on links quality. where a gateway residing at the root of the tree (Internet access gateway) frequently requests routes to other nodes of the network. 0.9 accumulated serving rate 0.. In particular. SrcRR.A. 22. [36] discuss the benefits of using SDN to decouple the operation and management of rural wireless networks from the physical infrastructure.A. authors in [52] investigate and evaluate the SDN control plane distribution for WMNs with emphasis on master controller selection. since the WMN nodes do not support OpenFlow). Kabel Deutschland (KD) [17].21. is another WMN routing protocol developed within the Freifunk community [6] to address the scalability limitations of the OLSR protocol and account for the static nature of WMNs. Dimogerontakis et al. Number of re-assigned flows with different topologies for 60 h simulation time.62] present OpenFlow-based architectures for WMNs with emphasis on mobility management.g. In our experimental evaluation. [38]) where routes are computed in advance leading to lower response time at the cost of high routing control overhead.4 0.A. This can complement our work to provide higher resilience and SDN controller availability. Furthermore. Our approach of coupling WMN routing protocols with SDN is also employed by authors in [28. proactive and reactive) are combined to adjust the tradeoff between response time and control overhead. All these protocols can be integrated into our SDN architecture for the selection of WMN paths between the home routers or as a fallback in case the connection between a router and its controller is lost. 369 Routing in WMN.A.6 0. LQSR [32] and SrcRR [19]) where both approaches (i. These techniques are complimentary to our work and can be integrated into our SDN control plane to facilitate mobility management in crowd-shared WMNs.1 0 TFA (21) 50 100 150 200 topologies with different size (# of nodes) Fig.M. routing.29. scalability. 5000 number of re-assigned flows 4500 4000 3500 3000 2500 2000 1500 1000 500 0 TFA (21) 50 100 150 200 topologies with different size (# of nodes) Fig. Abujoda et al. AODV [47].M.M..2 0. 23. WMN routing protocols can be classified into: (i) proactive (e. Recent work has been leveraging on SDN for WMN management.A. [30] propose a SDN extension to WMN community networks for L2 experimentation. Our work is in the same direction. More specifically. B. DSR) which provide routes on demand leading to lower control overhead.5 0. AODV-ST. Software-defined wireless networks.4 0. There is a large body of work on routing in WMN [20. Other applications of SDN to the wireless network domain include OpenRoads [63] and OpenSDWN [59]. Over the years. Hasan et al. OLSR [27]. WRP [46]. and load balancing..1 0 0 10 20 30 40 50 60 time (hours) Fig.T. low response time. but higher response time.A.FON [5].N.g.62].N. but we focus on the management of crowd-shared WMN in residential areas and investigate techniques for the efficient utilization of the shared bandwidth.e.T. AODV-ST [50] performs routing assuming a tree-based traffic flow. / Computer Networks 93 (2015) 359–372 1 TFA (21 nodes) 50 nodes 100 nodes 150 nodes 200 nodes 0. Authors in [28.) have evolved to target particular challenges or structures of WMNs. This diversity stems from the different optimization goals of routing protocols (e.T.29.e.N.. WMN utilization with different topologies for 60 h simulation time..3 0.6 WMN utilization 0. QoS support)..g. we rely on the WMN for the computation of routes between the local and the remote gateway (i.2 0.5 0.A. and (iii) hybrid (e.g. (ii) reactive (e.

This. DT and BT have partnered with FON to provide Internet access through shared home broadband connections. In contrast. E.pdf (accessed 18.de/wlan-hotspots/homespot-service. Wang. These efforts have also been supported by several affordable. FON. Content caching and the ever-growing penetration of online social networks (OSN) create for more opportunities in software-defined crowd-shared networks.com/ (accessed 18.jsp (accessed 18.org/wk/index.15). In future work. as opposed to a crowd-shared network with a single point of access where a portion of the shared bandwidth is wasted. such as the controller placement. http://h17007. leveraging on the trust between two parties with an OSN relationship [26]. we plan to investigate any potential benefits by coupling content caching and OSNs with crowd-shared WMNs.11 Mesh Networks. HP [11]. OpenFlow has been ported to OpenWRT [9.hp. On the other hand.e. Akyildiz. html (accessed 18. 2014.telekom.03. 7. Wang. Our approach leverages on SDN to bring the following benefits to Internet access sharing: (i) reduction of operational expenses for home networks and ISPs. W. Aguayo. D. the presence of a single point of access for each guest.15).rice.03. in: IEEE International Workshop on Community Networks and Bottom-up-Broadband(CNBuB 2014).net/ (accessed 18.edu/ (accessed 18. http://freifunk.linksys.03. while the SDN controller can redirect guests to caches.15). BT and KD use Home Hub and homespot. Community Lab.e. CONFINE Project. such as packet reordering and control communication overhead. as all sharing policy and WMN utilization information can be conveyed to a centralized controller facilitating the assignment of guest flows to gateways and the traffic redirection configurations in the WMN. easy-to-install and openfirmware wireless routers offered by manufacturers such as FON [9].15). since the deployment.03. Furthermore. popular content can be placed in caches within home networks. We envision hosting the SDN controller as a virtual machine in micro-datacenters hosted by large ISPs [35]. Meru [12]). Larnaca.03. A crowd-shared WMN can mitigate the fundamental problem of any crowdshared network. A survey on routing algorithms for wireless ad-hoc and mesh networks. Software-defined crowd-shared wireless mesh networks. virtual network operators).52]. We achieve the same goal (i. Wireless mesh networks: a survey.. configuration. due to the global WMN view and the collection of all monitoring and sharing policy information in a centralized controller. http://www. control plane distribution for robustness and load balancing. Abujoda et al. https://community-lab. By deploying the controller in micro-DCs we ensure low communication delay with the home routers and scalable processing power to cope with high loads (scale up the controller computation capacity by using more servers). Acknowledgments This work was partially supported by the EU CONFINE Project (288535). Papadimitriou. R.03.net (accessed 18. i. B.15). such that home users have access to higher upstream capacity. Furthermore. enabling resource pooling from all shared broadband connections. Furthermore. Freifunk. allowing wireless routers running OpenWRT to be configured using OpenFlow.pdf (accessed 18.php/Pantou_:_OpenFlow_1. leading to faster response time and providing a fallback controller in case of failure [22.15). A.03. Deploying a SDWMN in the real world raises challenges.awmn. We showed that the advance knowledge of user sharing policies can reduce the number of guest flow redirections (particularly for low and moderate network loads) avoiding implications.kabeldeutschland. Abujoda. 56 (2012) 940–965. Netw.openflow.edu/rtm/ srcrr-draft.mil/itd/ncs/products/mgen (accessed 18. PAWS) by connecting the home routers as a mesh. !http://pdos.03. http://www. 2003. Sathiaseelan. and Netgear [14].31.pdf? I. SDN brings significant benefits to crowd-shared networks. in turn. Comput. especially at the event of message loss. Mukherjee.com/help/about-bt-fon. since PAWS already uses wireless routers with OpenWRT-based firmware. http://www.btwifi.15). Linksys [13]. J. 47 (4) (2005) 445–487.com/collateral/white-papers/ sdn-for-wifi-wp.10]. http://corp. This essentially facilitates the extension of PAWS as a crowd-shared SDWMN. openMesh [15].23.eu/ (accessed 18.42. 0_for_OpenWRT (accessed 18.15).com/media/consumer-products/180150 (accessed 18. http://tfa. SrcRR: A High-Throughput Routing Protocol for 802.com/gb/c/wireless-routers/ (accessed 18.merunetworks. http://archive. Alotaibi. Conclusions In this paper.com/docs/campus-lan/c04219571-hires.nrl..com/ (accessed 18. there are already OpenFlow-based wireless routers (e. enables a software-defined crowd-shared WMN to accommodate a substantially larger volume of guest traffic.g. .open-mesh. Our experimental results show that our SDWMN control plane exhibits low per-flow setup time and low control communication overhead. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] http://www.openwrt.. TFA Rice. We can further leverage on several frameworks to distribute the control load across multiple servers.03. For example.. http://www. Micro-data centers are small DCs deployed at POP and network aggregation points.com/gb/c/wireless-routers/ (accessed 18.15).15). Cyprus. for Internet access sharing with other subscribers.03.mit.03.03.15).net/ (accessed 18. This leads to a significantly higher utilization of the shared bandwidth.15). and operation of crowd-shared networks can be outsourced to third-parties (i. More specifically. https://www. respectively.g.csail.www1. Rizk.03. Link-alike [40] promotes sharing by aggregating the uplink bandwidth of the broadband connections through WMN. high utilization of the shared bandwidth) via guest traffic redirections through the WMN. A. http://confine-project.15). / Computer Networks 93 (2015) 359–372 of shared hotspots.03.03.15).03. http://wiki.15).navy. ARBOR [61] aggregates broadband connection bandwidth (both uplink and downlink) from multiple WiFi networks.e. Netw.fon.org/toh/start (accessed 18. In terms of SDN support in home networks. Morris. http://www. we investigated the benefits of extending the coverage of any crowd-shared network (e. X. information can be utilized from OSNs in order to associate and authenticate guests to access points that belong to their online friends. A. a decentralized crowd-shared WMN management would require synchronization primitives and could yield slow convergence. Bicket.03.linksys. Comput.. In addition. conserving bandwidth in shared broadband connections. and (ii) better coordination of network configurations for traffic redirections. http://www. http://www.15). We thank Amr Rizk for his help in modeling the home router on/off periods.370 A. and SDN support SDN in home routers.15). P.

C. [23] Z.2470.. Estoril. Brewer. S. USA. P. Eidson. in: Proceedings of the Passive and Active Measurements (PAM) Workshop. Extending networking into the virtualization layer. Kohler. Itai. B. M. ACM SIGCOMM CCR 38 (2) (2008). T.E. A new end-to-end probing and analysis method for estimating bandwidth bottlenecks. in: Proceedings of the IEEE/IFIP SDNMO 2014. [42] T. NPART . J. N. Papadimitriou. D. Even. Draves. in: Digital Economy All Hands Meeting. P. Bozakov. Y. USA.. CoNEXT Student 12. T. C. [26] Z.IEEE [62] F. A. Uhlig. Towards an elastic distributed SDN controller. His research interests include network virtualization and network management with focus on algorithms and optimization. J. [27] T. LCD-Net: lowest cost denominator networking. M. Saldana. Milic. Malek. A. Fitschen. Int. [39] M. et al. 114–128. G. Papadimitriou. in 2006. Feldmann. Papadimitriou. 2009. N. Crowcroft. Melander. K.O. IETF. Smaragdakis. Amidon. S. Kassler.. Cao. in: Proceedings of the Annual ACM Special Interest Group on Data Communications (SIGCOMM94). [24] B. S. pp. Papadimitriou.. Braem. E. J. 2010. Zhu. 2012.L. degree in electrical and computer engineering at the Institute for Communications Technology at the Leibniz Universitt Hannover. F. [59] J. M. Radzik. org/abs/1406. HotSDN. Scott. [35] B. Syst. et al.E. Dixit.uk/news/technology-11162656. Sathiaseelan. S. P. in: Proceedings of the ACM International Conference on Mobile Computing and Networking (MobiCom). He also obtained a master of science degree in Mechatronics engineering from University of Siegen. 234–244. Schulz-Zander. Murthy. Pisa. Blefari-Melazzi. p. Ntlatlapa. [32] R. S. S. Koponen. Casado. CA. Controller selection in a Wireless Mesh SDN under network partitioning and merging scenarios. Commun. T. Garcia-Luna-Aveces. Commun. [49] B. / Computer Networks 93 (2015) 359–372 [22] Z. in: MC2R. A. Wireless Mesh software defined networks (wmSDN). Pfaff. in: Proceedings of the IEEE European Workshop on Software Defined Networks EWSDN. Bozakov. [61] X. in: Proceedings of the IEEE Consumer Communications and Networking Conference. Bjorkman. N. Berlin. Local Econ. pp. C. ACM SIGCOMM CCR 43 (2) (2013). Internet Draft. [38] D. et al. Shenker. [40] S. [60] L. New York. [41] E. 2008. M. Sriram.New York. Miller. 2015. Crowcroft. in: Proceedings of the CNBuB Workshop at IEEE WiMob. Dovrolis. Schneider. P. Simple pragmatic approach to mesh routing using batman. Dely. Seshan. Jacquet. Routing in multi-radio. USA. Buddhikot. Salsano. A. in: ACM HotNets. Germany. South Africa. 2004. 2013. 2012. 1 (2) (1996) 183– 197. Feldmann. Highly dynamic destination sequenced distance-vector routing (DSDV) for mobile computers. ACM SIGCOMM ebook on Recent Advances in Networking. Internet on the move: challenges and solutions. ACM Trans. [29] A. Haddadi.Wiley [58] A. Link-alike: using wireless to share network resources in a neighborhood. Siracusano. J. 2003. [56] A. Navarro. in: Proceedings of the IEEE INFOCOM. Reorder detecting TCP. G. Comput. in: Proceedings of the Annual Symposium on Computing for Development ACM DEV. [51] J. M. [53] A. he has been working towards the Ph. Krakow. P. OpenFlow: enabling innovation in campus networks. A. F. Germany. Steenkiste. 1994. Germany. Andersen. Schmid.D. Fildes. L. Hama. R. Evaluation and characterization of available bandwidth probing techniques. C. Y. Perkins. Hasan. in: Proceedings of the IEEE ICCCN. Lakshman. he has been working towards the Ph. pp. Public access WiFi service (PAWS). Zill. Sathiaseelan. G. S. Detti. Portugal. et al. [43] N. M. 2014. Vilata.large-scale production networks.bbc. E. I. [46] S. Aberdeen. Belding-Royer. Virtual public networks. J. in: Proceedings of the IEEE PIMRC. Hu. CCNC. [25] M. Detti. N. Jahnotti. 1–9. Poutievski. Maggs. et al. 5 (1976) 691–703. Wang. Casado. Ben-David. in: RFC 3561 (Experimental)„ IETF. Yang. Belding-Royer. J. C. OpenSDWN: programmatic control over home and enterprise WiFi. Shamir. in: Proceedings of the ACM SIGCOMM SOSR. Mishra. Ad hoc on-demand distance vector (AODV) routing. Chen. A Case for Research with and on community networks. 34. 19 (6) (2005). B. Poland. in: Proceedings of the ACM SIGCOMM Workshop on Hot Topics in Software Defined Networking. in 2009. Morris. [36] S. M. http://www. Hallstrom. J. D. Kaminsky. Ventre. the EUfunded T-NOVA project and the national project G-Lab. S. SIAM J. in: Proceedings of the IEEE International Conference on HSNMC. B. Comput. OpenRoads: empowering research in mobile networks. Iwata. in: Proceedings of the IEEE Workshop on Wireless Mesh Networks (WiMesh). Ramachandran. Reorder notifying TCP (RN-TCP) with explicit packet drop notification (EPDN). P. His research interests include software-defined networking. Kasshoek. R.). China. I. 2013. [52] S. Design and Implementation. Ahmed Abujoda is a student member of IEEE. G. P. [28] P. in: Proceedings of the IEEE JSAC. 2011. T. Sathiaseelan. Aichele. Digital Futures. David Dietrich is a student member of IEEE and ACM. J. 2015. in: Proceedings of the 2012 ACM Conference on CoNEXT Student Workshop. multi-hop wireless mesh networks. 22 (1) (2008) 6–12. Xing. [57] A. Blefari-Melazzi. ACM SIGCOMM CCR 43 (3) (2013). V. X. Shenker. Frank. Openflow for wireless mesh networks. J. Sathiaseelan. Since 2010. CSIR. Clausen. Routing metrics and protocols for wireless mesh networks. submitted paper. [31] A. [33] S. USENIX association. Chandranmenon. 305–377. network function virtualization and network management. J. Sathiaseelan. [55] A. e. S. in: Operating Systems. David Dietrich graduated in electrical engineering from the Universitt Kassel. Syst. Inouye. O. Dimogerontakis. Spec. During the years 2006 until 2009 he was a software engineer for real-time simulation of automotive communications networks.M. France. L. Gunningberg. Las Vegas. Almeroth. Jakubczak. 2013. [54] A. Gondi. Aggarwal. 18 (3) (2000). Santa Clara. Netw. Ahmed Abujoda graduated in electrical engineering from the university of Sana’a. Bonaventure (Eds. P. IEEE. Papagiannaki. G. 2013. Hong Kong. T. Yap.C. Sathiaseelan. S. V. Padhye. Community Networks: Definition and taxonomy. [45] B. in: Proceedings of the IEEE WiMob. Maui. On the design and implementation of infrastructure mesh networks. Gude. Fairhurst. Perkins.M. Abujoda et al. 28 (6) (2013). Bhagwat. Pettit. Enhanced broadband access as a solution to the social and economic problems of the rural digital divide. [44] B. 7. 2010. NY. [63] K. Collaboration opportunities for content delivery and network infrastructures. McKeown. Mukherjee. E. A. Towards a scalable software-defined network virtualization platform. Stribling. He also participated in several research projects. Yemen. in: H. [37] N. P.M. in: Proceedings of the 2nd International Conference on Simulation Tools and Techniques. Social WiFi: hotspot sharing with online friends. Das. USA. ACM SIGCOMM CCR 43 (1) (2013). Koponen. Pretoria. 2002. An efficient routing protocol for wireless networks.. Sathiaseelan. ACM. Pathload: A measurement tool for end-to-end available bandwidth.D. pp. On the complexity of timetable and multicommodity flow problems. in: Proceedings of the ACM SIGCOMM. M. Issue Routing Mob. S. Trossen. et al. B. 2003. BBC. 2008. Hao. USA. K. OpenFlow-based load balancing for wireless mesh infrastructure. 2003. Lyon. Shenker. Poese. 2005. C.A. [34] J. 2010. R. Crowcroft. Pisa. Salsano. C. Germany. 2009. ACM. K. Radzik.RFC 3626 (Experimental). C. .. 2009.N. in: Proceedings of the Global Internet Symposium. Campista. Germany. Barcelona. ACM/Baltzer Journal on Mobile Networks and Applications. Kompella. Onix: a distributed control platform for. New York.S. 2010. 6–7 October 2008. 2013. [47] C. Liu. 371 [48] C. Spain. N. Johnson.11 wifi networks. The Click Modular Router. NY. 2000. Bayer.J. available at http://arxiv. Rotsos. [50] K. Since 2010. C. Optimized Link State Routing Protocol (OLSR).. J. Jain. Wallace. 2013. ARBOR: hang together rather than hang separately in 802. 2003.co. Software defined networking for community network testbeds. UN reveals global disparity in broadband access. A. IEEE Netw. Ramanathan. in 2004.F. in: Proceedings of the 2nd IFIP International Symposium on Wireless Communications and Information Technology in Developing Countries. 2014. J.g. Ciobotaru. K. Enhancing rural connectivity with software defined networks. Mayer. degree in electrical and computer engineering at the Institute for Communications Technology at the Leibniz Universitt Hannover. H. Autoslice: automated and scalable slicing for software-defined networks.V. 2014.Node Placement Algorithm for Realistic Topologies in Wireless Multihop Network Simulation. [30] E. Townsend.

India (2000). Abujoda et al. He is the Chair of IRTF Global Access to the Internet for All (GAIA) research group and a member of the Internet Research Steering Group (IRSG). . He received a Ph. Greece. network function virtualization. France. in Networking from Kings College London (2005). He has a Ph. UK. in Computing and Internet Systems from Kings College London (2001) and Bachelors in Computer Science and Engineering from NIT. The research group conducts research on novel Internet architectures for improving and reducing the cost of Internet access. / Computer Networks 93 (2015) 359–372 Panagiotis Papadimitriou is an Assistant Professor at the Communications Technology Institute of Leibniz Universitt Hannover. in Electrical and Computer Engineering from Democritus University of Thrace.372 A. Arjuna Sathiaseelan is a Senior Research Associate at the Computer Laboratory. He leads the Networking for Development (N4D Lab). in 2008. and at Telecom SudParis.Sc. From 2008 to 2010.D. Trichy. and cloud networking. The research activities of his group currently span nextgeneration Internet architectures. software-defined networks.D. M. University of Cambridge. Panagiotis was a PostDoctoral Researcher at the Computing Department of Lancaster University. Panagiotis received the runnerup Poster Award at ACM SIGCOMM 2009 and the Best Paper Award at IFIP WWIC 2012.