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

a Institute of Communications Technology, Leibniz Universität Hannover, Germanyb 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

Keywords:

SDN

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 Less-

than-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 Notting-

ham, 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 In-

ternet 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).

http://dx.doi.org/10.1016/j.comnet.2015.09.008

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, be-

cause 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 economi-

cally motivated) to provide wired and cellular infrastructures

360 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

to rural/remote areas have led to several initiatives to build

large-scale, self-organized, and decentralized community

wireless networks that use WiFi mesh technology (includ-

ing long distance), due to the reduced cost of using the unli-

censed spectrum [51]. These community wireless mesh net-

works have self-sustainable business models, which provide

more localized communication services, as well as Internet

backhaul support via peering agreements with traditional

network operators who see such networks as a way to ex-

tend their reach at a lower cost. There are also community-

led wireless initiatives such as crowd-shared wireless net-

works, in which home broadband owners share a portion of

their home broadband with friends, neighbors, or other users

either for free or as part of a service offering by the ISP (e.g.,

[6,53]).

Public Access WiFi Service (PAWS) [53] is a community-

led crowd-shared WiFi service that uses a set of tech-

niques that make use of the available unused capacity in

home broadband networks and allowing Less-than-Best Ef-

fort (LBE) access to these resources, based on the Low-

est Cost Denominator Networking (LCDNet) paradigm [58].

PAWS adopts an approach of community-wide participation,

where home broadband subscribers are enabled to donate

controlled but free use of their high-speed broadband Inter-

net to fellow citizens. PAWS was deployed with 20 custom-

made PAWS routers placed in a deprived community in Not-

tingham and was also trialed out in rural Wales. PAWS is

essentially a crowd-shared access network (similar to FON

[5]) for under-privileged users in urban and rural communi-

ties. PAWS has been facing ongoing deployment challenges,

such as limited coverage, stemming from user sharing pat-

terns. In particular, during the PAWS trial deployment, it was

observed that home users did not share their broadband con-

nection over periods at which either the whole bandwidth or

all the ports of the home router were needed (i.e., PAWS uses

an access point connected to the home router for Internet ac-

cess sharing). Essentially, PAWS is a crowd-shared network

with a single point of access per guest, and as such, Internet

access sharing is highly dependent on user sharing policies

(i.e., the periods at which user share their Internet connec-

tion).

To mitigate this problem, we investigate the potential

benefits of extending PAWS or any crowd-shared wireless

network to a wireless mesh network (WMN) by intercon-

necting wireless home routers. As such, a crowd-shared

WMN provides extended coverage via multiple points of ac-

cess for each guest. We particularly consider crowd-shared

WMNs in residential areas, taking advantage of the dense de-

ployment of wireless home routers. The main challenge in

the management of such a WMN lies in the coordination of

guest traffic redirections, such that the shared bandwidth is

efficiently utilized. More precisely, 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. Furthermore, decisions for traffic redirections

should be also based on user sharing policies, when these are

disclosed in advance. Given the amount of information that

has to be collected before flow assignments can be made, we

deem a centralized control plane as a more suitable approach

to WMN management for Internet access sharing, since all

information can be conveyed to a centralized controller

facilitating the coordination of traffic redirections. In this re-

spect, we leverage on software-defined networking (SDN)

principles for the control plane design.

Along these lines, our contributions are the following:

• We present the design and implementation of a SDN con-

trol plane for the coordination of guest traffic redirections

through the WMN.

• We develop an algorithm for the assignment of gateways

to flows that require redirection, taking into account the

flow demands, the residual bandwidth in the access links

and the WMN, as well as, the advance knowledge of shar-

ing policies.

• We generate a model for user sharing patterns based on

the router on/off periods captured from the PAWS deploy-

ment in Nottingham.

• 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], i.e., one of the community networks of

the CONFINE project [2,3,24]. Community Lab, https://

community-lab.net/ In this respect, we perform a com-

parative study between a WMN and a crowd-shared net-

work with limited coverage (i.e., one access point per

guest).

This paper extends our previous study on software-

defined WMNs, at which the efficiency of a crowd-shared

WMN was assessed using simulations [18]. The remainder of

the paper is organized as follows. In Section 2, we provide

an overview of the software-defined WMN (SDWMN) con-

trol plane. Section 3 elaborates on techniques for guest traf-

fic redirection. In Section 4, we discuss the implementation

of the SDWMN and its deployment in AWMN. In Section 5,

we present our evaluation results and discuss the benefits of

a crowd-shared WMN for Internet access sharing. Section 6

discusses related work. Finally, Section 7 highlights our con-

clusions.

2. 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,

they have no provision to extend the coverage when no band-

width is being shared. Based on our experience from the trial

PAWS deployment, PAWS routers were not available for cer-

tain periods, because sharers needed all the bandwidth of

their broadband connection or due to other reasons, such as

economic constraints placed on home users in underprivi-

leged areas where they are enforced to conserve energy by

turning off the routers at nights. These observed user behav-

iors entail significant challenges for the successful adoption

of PAWS.

A potential solution to this problem is to extend the PAWS

network as a crowd-shared WMN. Such a network would

allow home network users to share part of their own broad-

band connection with the public for free while also con-

nected to each other as a WMN providing extended cover-

age (Fig. 1). Extending PAWS to a crowd-shared WMN departs

from the norm: multiple users from different ISPs form part

A. Abujoda et al. / Computer Networks 93 (2015) 359–372 361

Fig. 1. Crowd-shared WMN for public Internet access.

Sharer

Community Network

Traffic Redirection

Southbound API (OpenFlow, SNMP)

Monitoring

Northbound API

SD

N C

ont

rol P

lane

Virtual Network Operator

Research Device

Research Device

Research Device

Research Device

Research Device

Fig. 2. Software-defined WMN control plane overview.

of the WMN to provide free Internet connectivity, while most

wireless community WMNs today are operated by a single

organization. This raises important questions regarding the

operation, configuration, and management of crowd-shared

WMNs.

SDN can facilitate the management and operation of

wireless networks at large scale. Leveraging on SDN’s

centralized control and network-wide visibility, the man-

agement and operation of a crowd-shared WMN can be

outsourced to a third party. In [54], 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).

Compared to existing mesh networks for Internet access

sharing (e.g., Freifunk [6]), our approach provides more

opportunities for non-governmental organizations and local

governments (driven by social goals rather than economic)

to become virtual network operators. 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.

In particular, we rely on SDN to create the notion of

Virtual Public Networks (VPuN), i.e., crowd-shared home

networks created, deployed and managed through an evo-

lutionary SDN control abstraction [54]. Although originally

intended for crowd-shared wireless networks such as PAWS,

VPuN can be also used for crowd-shared WMNs, enabling

resource pooling across multiple home broadband connec-

tions based on the prevailing network conditions and usage

sharing patterns. Based on the notion of VPuN, we have

deployed a SDWMN control plane in AWMN [4]. As part

of the Community-Lab [3], AWMN consists of more than

1000 wireless nodes and 30 research devices (RDs) that

host isolated containers (i.e., so-called slivers) which can be

allocated and controlled by a user, according to the needs of

his experiment [24].

Fig. 2 illustrates an overview of the SDWMN control plane.

The main goal of the control plane is to improve the uti-

lization of shared bandwidth by redirecting guest traffic to

one of the home routers through the WMN, based on user

sharing policies. As such, the control plane exposes an inter-

face to each home user, allowing him to express a sharing

policy (e.g., using the sharing policy expression language

presented in [54]). The control plane functionality mainly

consists of traffic redirection and monitoring. The traffic

redirection module selects the gateway for Internet access

(using the algorithm presented in Section 3) and subse-

quently creates a tunnel between the home router where

the guest is connected and the router assigned as the gate-

way (home routers are deployed in the AWMN RDs). Traffic

is redirected through a tunnel, 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 redi-

rected over that path, and it is torn down when it is no longer

needed. The monitoring module collects statistics about the

bandwidth utilization of the WMN and the access links. A de-

tailed description of the implementation of the SDWMN con-

trol plane and the gateway is given in Section 4.

3. Traffic redirection

Assigning flows to network paths with constrained capac-

ity can be formulated as a multi-commodity flow problem

which is known to be NP-hard [33]. Hereby, we present a

heuristic algorithm for the assignment of the gateway and

the path over which the traffic will be redirected through

the WMN. The algorithm aims at maximizing Internet shared

bandwidth utilization by accommodating as large as possible

number of flows. It is executed by a SDN controller which has

knowledge of the WMN topology and utilization, the Internet

access link utilizations, and the user sharing policies. We first

describe the traffic redirection algorithm (Section 3.1) and

then investigate the potential benefits of user sharing poli-

cies (Section 3.2).

3.1. Traffic redirection algorithm

We represent the WMN as a weighted undirected graph

G = (N, L), where N is the set of nodes and L is the set of links

362 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

Algorithm 1 Gateway and path assignment.

Inputs: G = (N, L), 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 ← ng // gateway assignment

path ← SP(Pug) // path assignment

search ← false

end if

end while

end for

C

Fig. 3. Flow redirections.

between nodes of the set N. Each node ni is associated with

an Internet access link whose available shared bandwidth is

denoted by C(ni). Each link lij ∈ L between two nodes ni and

nj is associated with the available bandwidth C(lij). Let Pij de-

note the set of paths in the network G, between the pair of

nodes ni and nj. The available bandwidth C(p) associated to a

path p ∈ Pij is given by the minimum residual bandwidth of

the links along the path:

(p) = minli j∈p

C(li j) (1)

We further represent a flow demand for a guest with the

tuple d = (nu, r), where nu ∈ N is the node where the guest

device has been attached and r denotes the flow rate. We use

D to represent the set of flow demands that have arrived in

the network.

Algorithm 1 assigns a gateway and the shortest path to

each flow demand from the set D. The algorithm is executed

whenever there is insufficient shared bandwidth in the lo-

cal home network. Initially, the flow demands are sorted in

decreasing order. For each flow demand, the algorithm se-

lects the home router with the highest access link bandwidth

(i.e., C(ni)). Subsequently, the algorithm identifies the set of

paths, Pug, between the local home router (nu) and the as-

signed gateway (ng) that satisfy the flow demand (Eq. (1)). In

case there is no such path, the algorithm performs another

iteration for the selection of the gateway, excluding the pre-

viously selected home router. Otherwise, the shortest among

these paths is being identified based on the number of hops

(SP function in the algorithm). This eventually computes the

path for the traffic redirection through the WMN.

This algorithm carries out the gateway assignment based

on the available shared bandwidth of the home routers in

the crowd-shared network. Although this can pool resources

from all home networks where sharing is permitted achiev-

ing efficient utilization of the shared bandwidth, guest traffic

may have to traverse multiple hops in the WMN. This can

lead to latency inflation, delay variation, reordering [56,57]

and in general, performance degradation. To mitigate this,

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. For example, adjusting the

threshold to 1 restricts the search space to the next-hop

routers with sufficient shared bandwidth.

3.2. User sharing policies

Algorithm 1 assigns flows to gateways taking into ac-

count the amount of shared bandwidth and optionally the

path hop-count. So far, we have not exploited any advance

knowledge of user sharing policies for traffic redirection.

Such information can prevent wasteful traffic redirections

(i.e., forwarding traffic to a router which will shortly become

unavailable for Internet access sharing). Sharing policies

can be specified using a sharing policy expression language

(SPEL), e.g., the XML-based schema presented in our previous

work [54]. SPEL essentially provides the necessary means to

the sharers to specify the amount of shared bandwidth as

well as the period of sharing. This information is conveyed

to the SDN control plane and is used as an additional input

for gateway assignment.

We run a trace-driven simulation of a crowd-shared

WMN to investigate the impact of sharing policies on traf-

fic redirections. In particular, we model the availability of

the home routers using an on-off Markov chain. On and off

times are exponentially distributed with mean values μon =106 min and μof f = 555 min. We parameterize the exponen-

tial distributions using datasets from the PAWS deployment

in UK.

Fig. 3 illustrates the number of traffic redirections for dif-

ferent network loads with and without the knowledge of

sharing policies. When sharing policies are known, the traffic

A. Abujoda et al. / Computer Networks 93 (2015) 359–372 363

path 1

Sink

delay node (tc)

OVSSrc

path 2

Fig. 4. Experimental setup for packet reordering evaluation.

0 100 200 300 400 5000

25

50

75

100

125

150

175

200

225

delay(old_path) − delay(new_path) (ms)

num

ber

of p

acke

ts o

ut−

of−

orde

r

1 Mbps2 Mbps5 Mbps

Fig. 5. Packet reordering.

Fig. 6. SDWMN controller.

is directed to the router with the longest period of availability

(among the routers with sufficient shared bandwidth). Ac-

cording to Fig. 3, using sharing policies leads to fewer traffic

redirections, especially for low and moderate network loads.

For higher loads, the number of routers with sufficient shared

bandwidth diminishes, and as such, the knowledge of sharing

policies does not yield significant gains.

We perform additional tests to investigate potential im-

plications, such as packet reordering and higher communi-

cation overhead, due to traffic redirections. First, we use the

experimental setup depicted in Fig. 4 to quantify the level of

packet reordering. In this setup, UDP traffic is redirected from

path 1 to path 2 at the intermediate node. The redirection is

carried out by inserting flow entries to an OpenvSwitch (OVS)

datapath that has been set up in the intermediate node. We

further employ a delay node, set up in path 2, to adjust the

delay along this path. Fig. 5 shows that the number of out-

of-order packets increases in proportion to the delay differ-

ence between the two paths. Our path delay measurements

in AWMN indicate that a redirection through the WMN in-

creases the RTT with up to 50 ms. This will result in the

out-of-order delivery of a small number of packets, which, in

turn, can lead to congestion control invocations, and, hence,

lower throughput for TCP flows.

Besides packet reordering, traffic redirections generate

additional communication overhead between the SDWMN

controller and the routers. Specifically, for each flow redirec-

tion the router sends a OFPT_PACKET_IN message to the con-

troller, which, in turn, sends an OFPT_FLOW_MOD message to

the router. Taking the respective acknowledgment messages

into account, the communication overhead per flow redirec-

tion is 490 bytes. This overhead can increase substantially

with a large number of guest flows per home router, as the

PAWS router logs indicate. Considering the identified impli-

cations of traffic redirections, we use sharing policies in the

assignment of guest flows to gateways.

4. Implementation

We provide a detailed description of the implementation

of the SDWMN control plane and gateway in Sections 4.1 and

4.2, respectively.

4.1. SDWMN controller

We implement our controller in Python using POX which

provides a framework to implement SDN controllers with

OpenFlow protocol interface [43]. Since OpenFlow supports

only the configuration of switch flow tables, we use XML-RPC

interface to install and configure tunnels between the gate-

ways. Our SDWMN controller consists of the following mod-

ules (Fig. 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. Each

gateway object holds information about the gateway’s:

(i) datapath ID, which is a 64 bit unique identifier for

each OpenFlow switch instance, (ii) IP address of the

WMN port, (iii) ports table, which contains the names and

OpenFlow switch ports numbers (i.e., DSL port, the WMN

port and the guests’ wireless network port), (iv) moni-

toring table, which contains the gateway shared band-

width utilization and WMN path quality, and (v) tunnel

table, which contains the IP addresses, UDP port numbers

(we use UDP-in-IP tunnelling) and OpenFlow switch port

numbers of the installed tunnels. The registration process

is triggered by POX when the gateway establishes a new

control channel with the controller.

• Monitoring: This module collects and processes the mea-

surements of shared bandwidth utilization and the WMN

paths delay and hop counts for each gateway. Bandwidth

(BW) estimation in a WMN is very challenging, due to BW

fluctuations and the low degree of accuracy of available

BW measurement tools (e.g., pathload [39], TOPP [44],

and IGI/PTR [37]) that was also confirmed by our own

tests. However, in our experimental setup, the Internet

access links are the bottleneck (since they have been con-

figured at lower capacity than the WMN), and, as such,

364 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

Fig. 7. SDWMN gateway. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

the aforementioned WMN bandwidth estimation issues

do not affect the accuracy of our measurements. Using

the OpenFlow protocol, 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.g., the DSL port). In particular,

the controller sends an OpenFlow OFPT_STATS_REQUEST

message with OFPST_PORT type and the port number to

the gateway. The gateway replies with OFPT_STATS_REPLY

message which carries the latest reads of the gateway

rx_bytes and tx_bytes counters. To calculate the shared

bandwidth utilization, the BW measurement module ap-

plies exponential moving average to the counters changes

over time. In addition, the path measurement module

uses XMP-RPC to pull the RTT and hop-count measure-

ments of each WMN path to/from each gateway. All mea-

surements are stored in the monitoring table of the re-

spective gateway object.

• Traffic redirection: Flows are redirected using the follow-

ing modules:

• Gateway selection: The gateway selection module is in-

voked when a gateway sends a OFPT_PACKET_IN mes-

sage to the controller to announce the arrival of a

new flow. Based on the measurements (shared BW

utilization and path length) provided by the moni-

toring module and using the algorithm described in

Section 3, the gateway selection module assigns a

gateway to the new flow. Subsequently, a tunnel is

being set up (see tunnel installation for further de-

tails) between the home gateway (where the flow ar-

rives) and the assigned gateway (where the flow is

redirected). Next, 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.

• Tunnel installation: Using the XML-RPC interface,

packet encapsulation modules are installed in the

gateway. 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. Upon successful

installation of the tunneling module, the home gate-

way replies with the switch port number of the new

encapsulation module. This information is stored in

the tunnel table of the home gateway object to be used

later by the flow table configuration module.

• Flow table configuration: This module installs the cor-

responding flow entries in the home gateway and

the assigned gateway using OpenFlow. This consists

in sending a OFPT_FLOW_MOD message to the home

and assigned gateways. This message carries the flow

matching fields [43] (e.g., source/destination IP, port

source/destination) as well as the OpenFlow switch

output port number.

4.2. SDWMN gateway

Our SDN gateway exposes to the SDWMN controller an

OpenFlow and XML-RPC interface to install tunnels and redi-

rect flows (see Fig. 7). Tunnels are created by encapsulating

guest traffic in UDP packets (i.e., IP-in-UDP) using the encap-

sulation module (red line Fig. 7). The destination IP address

of the tunnel header is set to the IP address of the assigned

gateway. For each new tunnel a new encapsulation module is

created. On the other hand, incoming traffic (green line Fig. 7)

is processed by the decapsulation module, which strips the

A. Abujoda et al. / Computer Networks 93 (2015) 359–372 365

Fig. 8. The WMN topology measured using traceroute from a research de-

vice (blue circle) toward the other research devices (red circles). (For inter-

pretation of the references to color in this figure legend, the reader is re-

ferred to the web version of this article.)

outer header before the traffic is delivered to the output port.

Both encapsulation and decapsulation modules are imple-

mented using Click [41].

To steer data traffic between the physical ports and the

modules (encapsulation/decapsulation), we rely on Open-

vSwitch [49]. This switching module also forwards messages

from the SDN controller to the gateway controller module

through the XML-RPC server. To connect modules to Open-

vSwitch, we use Linux TAP devices.

The instantiation of encapsulation/decapsulation mod-

ules is performed by the gateway controller in the following

steps: (i) creation of TAP interfaces, (ii) appending TAP in-

terfaces names, 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) in-

stallation of the Click configuration.

Furthermore, the gateway controller collects and trans-

fers the measurements generated by the path measurement

module to the SDN controller. The path measurement mod-

ule uses ping and traceroute to collect RTT and hop counts of

the WMN paths. A list of gateways IP addresses is passed to

this module to obtain the measurements. The module sends

10 probing packets every 30 s to avoid flooding the network.

5. Evaluation

In this section, we perform a comparative study of crowd-

shared WMN against PAWS (or any other crowd-shared net-

work with a single point of access). To this end, we run exper-

iments in AWMN (Section 5.1) and further use simulations

with larger WMN topologies (Section 5.2).

Using experimentation and simulations, we initially mea-

sure the utilization level of the shared BW and the accumu-

lated serving rate across time and for different flow arrival

rates. To study the scalability of our control plane, using ex-

perimentation we quantify the control communication over-

head in terms of BW consumption and flow setup delay. We

further use simulations to study the effect of traffic redi-

rection on latency by measuring the shared BW utilization

when applying a threshold to the redirection path length (see

Section 3). Finally, we investigate the scalability of the crowd-

shared WMN in terms of BW utilization and serving rate with

different network sizes using simulations.

5.1. Experimental results

For our experiments, we use 12 research devices (RDs) to

deploy 11 home routers (or gateways) and one SDWMN con-

troller. Fig. 8 illustrates the WMN topology measured from a

single RD. We implemented a tool to analyze and construct

the topology out of traceroute measurements. Using MGEN

[1], we generate guest flows with a rate and lifetime sam-

pled out of a uniform and an exponential distribution, re-

spectively. The flows arrive to the network according to a

Poisson process. For router availability pattern, we use the

model described in Section 3.2. Since running tests for more

than 555 min (the mean value of the off period) is not fea-

sible, we scale down the mean values to μon = 10.6 s and

μof f = 55.5 s. In each gateway, we further implement a timer

to emulate the on and the off period of the router. Upon the

expiration of the timer, the gateway sends a signal to the

controller to announce the beginning of an on or off period.

The time required to generate the signal is negligible (a few

milliseconds). This timer can be also used by the sharers for

sharing policy configurations in their home routers. Due to

occasional reliability issues in the RDs, we restricted the du-

ration of our experiments to 250 s. After downscaling the

router on/off periods, there are several transitions between

the router states during each experiment.

Our main evaluation metrics include: (i) the shared band-

width utilization across all home routers, and (ii) the accu-

mulated serving rate ASR(T) at time T, defined as:

ASR(T) = �Tt sfinished

t(�T

t sfinishedt + �T

t srejectedt

) , (2)

where sf inishedt denotes the total sizes of the flows at time t

which are accepted (i.e., assigned Internet access) and suc-

cessfully served without disruption due to router unavailabil-

ity. sre jectedt denotes the total sizes of the flows at time t which

are rejected (i.e., not assigned Internet access).

We initially measure the shared bandwidth utilization

(Fig. 9) and the serving rate (Fig. 10) with an arrival rate of

60 flows per minute, across 10 runs. Variations in the mea-

surements across the different runs were insignificant. Fig. 9

illustrates a low utilization of the shared bandwidth with-

out a WMN during the whole period, although there is high

demand for Internet access by guests attached to the vari-

ous home networks. In contrast, a WMN allows to capital-

ize the unused capacity and accommodate a larger volume of

guest traffic. More precisely, according to Fig. 9 guest traffic

366 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

Fig. 9. Shared bandwidth utilization.

Fig. 10. Accumulated serving rate.

Fig. 11. Shared bandwidth utilization for diverse flow arrival rates across

250 s experimentation time.

Fig. 12. Accumulated serving rate for diverse arrival rates across 250 s ex-

perimentation time.

redirection through the WMN results in the full utilization

of the shared bandwidth. Furthermore, crowd-shared WMNs

can accommodate substantially larger volume of guest traf-

fic, as depicted in Fig. 10. This stems from the high utilization

of the shared bandwidth.

We further measure the shared bandwidth utilization and

serving rate with a wide range of guest traffic demands. In

this respect, the boxplots in Figs. 11 and 12 illustrate the

shared bandwidth utilization and serving rate with diverse

flow arrival rates, ranging from 20 to 80 flows per minute.1

These results corroborate the efficiency of the WMN for vari-

ous traffic loads, as the shared bandwidth utilization and the

1 All boxplots show measurements across the time period of the exper-

iment, i.e., each point on the graph represents a different time point. For

different runs we calculate the average, the variation is insignificant.

serving rate always remain very high. In Fig. 12, the flow ar-

rival rates of 60 flows per minute and beyond lead to request

rejections, due to the insufficient access link bandwidth. On

the other hand, Figs. 11 and 12 show poor bandwidth utiliza-

tion and serving rate without a WMN, due to the presence

of a single point of Internet access for each guest. Essentially,

our results show the significant benefit that a WMN can bring

into crowd-shared networks, by effectively pooling resources

across all home networks.

To evaluate the scalability of our control plane, we mea-

sure the flow setup time and the average control communi-

cation overhead across a range of flow arrival rates. 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.

This is essentially the time incurred for gateway selection

and flow entry installation. We also define the control com-

munication overhead as the bandwidth consumed to setup

flows and monitor the shared bandwidth utilization. We run

A. Abujoda et al. / Computer Networks 93 (2015) 359–372 367

Fig. 13. Setup time per flow.

Fig. 14. Control communication overhead.

Fig. 15. Simulation WMN topology.

time (hours)0 5 10 15 20 25 30 35 40 45 50 55 60

shar

ed b

andw

idth

util

izat

ion

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

with WMNwithout WMN

Fig. 16. Shared bandwidth utilization.

our experiments on a single gateway (the results can be eas-

ily extrapolated for more gateways) and generate 100 differ-

ent flows for each arrival rate. Figs. 13 and 14 show that per

flow setup time does not change significantly, while the con-

trol communication overhead increases linearly and is in the

range of tenths of Kbps. According to these results, the perfor-

mance of our controller scales with increasing control loads.

5.2. Simulation results

To run tests at larger scale, we developed a simulator (in

Python) that models the flow-level behavior of the guest traf-

fic in a crowd-shared WMN. The flow parameters that we use

in our simulations are the average flow rate, and their arrival

and departure time. We use the TFA wireless mesh topology

[7] which consists of 21 nodes (Fig. 15). TFA is an operational

network deployed in a densely populated residential area;

that is exactly the network environment we consider for the

proposed crowd-shared WMN. Each node in this topology

represents a home router with shared Internet access band-

width, whereas each edge is a wireless mesh link. Guest flows

arrive randomly at different home routers. Each generated

flow has a rate and lifetime sampled out of a uniform and ex-

ponential distribution, respectively. The guest flows arrive to

the network according to a Poisson process. We simulate the

router availability using our model with the mean values ob-

tained from the PAWS datasets. Guest flows are granted Inter-

net access either through local routers or by redirection to re-

mote routers based on the algorithm presented in Section 3.

Flows which cannot be accommodated are rejected.

Each simulation run comprises 60 h at a time discretiza-

tion of 20 min. For each scenario, we perform 20 simulation

runs. We assume a homogeneous setting where each router

has 16 Mbps ADSL downlink and set the shared bandwidth

per router to 8 Mbps (i.e., in the periods that each router is

active). We run our simulation with mesh link capacity of

200, 54 and 10 Mbps. This does not have any impact on the

simulation results, since the bottleneck is the Internet access

links. Note that our wireless link model does not consider the

impact of interference and distance on the offered link band-

width.

The simulation results are in accordance with the results

from the community network. According to Figs. 16 and 17,

the crowd-shared WMN yields significantly higher utiliza-

tion of the shared bandwidth and accommodates more traffic

in comparison to the crowd-shared network without WMN.

368 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

time (hours)0 5 10 15 20 25 30 35 40 45 50 55 60

accu

mul

ated

ser

ving

rat

e

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

with WMNwithout WMN

Fig. 17. Accumulated serving rate.

flows arrrival rate (flows/min)10 20 30 40 50 60 70 80 90 100

shar

ed b

andw

idth

util

izat

ion

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

with WMNwithout WMN

Fig. 18. Shared bandwidth utilization vs. flow arrival rate across 60 h simu-

lation time.

flows arrival rate (flows/min)20 50 80

shar

ed b

anw

idth

util

izat

ion

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

max number of hops=1max number of hops=2max number of hops=3no limit

Fig. 19. Shared bandwidth utilization for diverse hop-count threshold val-

ues vs. flow arrival rate.

TFA (21) 50 100 150 200

shar

ed b

andw

idth

util

izat

ion

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

topologies with different size (# of nodes)

with WMN

without WMN

Fig. 20. Shared bandwidth utilization with different topologies for 60 h sim-

ulation time.

The serving rate, in particular, drops as guest flows are not

granted Internet access, due to either insufficient bandwidth

(with WMN) or changes in sharing policies (without WMN).

Eventually, in both cases, the serving rate reaches a steady

state. Shared bandwidth utilization for a wide range of guest

traffic demands is depicted in Fig. 18.

In our simulations, redirected guest traffic traverses up to

4 hops in the WMN. We further use a variant of our gateway

assignment algorithm (briefly discussed in Section 3), which

introduces a threshold in the number of hops between the

home router and the assigned gateway. Fig. 19 illustrates the

shared bandwidth utilization for diverse hop-count thresh-

old values ranging from 1 to 3, and without any limit in the

hop count. Restricting the number of hops has a noticeable

impact on bandwidth utilization, especially for a single hop,

since Internet access is permitted only through one of the

next-hop home routers. In essence, there is a trade-off be-

tween coverage extension (and thus effective bandwidth uti-

lization) and latency. This may become more critical in large

WMNs, where long paths can inflate latency.

We further run simulations using synthetic WMN topolo-

gies of diverse size and structure generated using NPART

[45]. NPART is a tool for generating realistic WMN topologies,

based on real-world WMN deployments (Berlin Freifunk[6]).

Using NPART and the Berlin WMN model, we randomly gen-

erate 4 additional WMN topologies with 50, 100, 150 and 200

nodes. We set the wireless link capacity and shared band-

width to 54 Mbps and 8 Mbps, respectively. For all topolo-

gies, we generate flows with arrival rate of 120 flows per

minute, which allows us to saturate the WMN capacity with

all topologies. As shown in Figs. 20 and 21, the shared band-

width utilization and serving rate scale linearly with the in-

crease of the WMN size. This is further confirmed by the

increase in WMN utilization (Fig. 22) and the number of

re-assigned flows (Fig. 23) with larger WMN sizes.

6. Related work

In the following, we discuss related work on WMN rout-

ing, software-defined wireless networks, and Internet access

sharing.

A. Abujoda et al. / Computer Networks 93 (2015) 359–372 369

time (hours)0 10 20 30 40 50 60

accu

mul

ated

ser

ving

rat

e

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1TFA (21 nodes)50 nodes 100 nodes 150 nodes 200 nodes

Fig. 21. Accumulated serving rate with different topologies.

topologies with different size (# of nodes)TFA (21) 50 100 150 200

WM

N u

tiliz

atio

n

0

0.1

0.2

0.3

0.4

0.5

0.6

Fig. 22. WMN utilization with different topologies for 60 h simulation time.

topologies with different size (# of nodes)TFA (21) 50 100 150 200

num

ber

of r

e-as

sign

ed fl

ows

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

Fig. 23. Number of re-assigned flows with different topologies for 60 h sim-

ulation time.

Routing in WMN. There is a large body of work on rout-

ing in WMN [20,21,25]. This diversity stems from the differ-

ent optimization goals of routing protocols (e.g., low response

time, low control overhead, scalability, QoS support). More

specifically, WMN routing protocols can be classified into: (i)

proactive (e.g., OLSR [27], WRP [46], DSDV [48], B.A.T.M.A.N.

[38]) where routes are computed in advance leading to lower

response time at the cost of high routing control overhead,

(ii) reactive (e.g., AODV [47], DSR) which provide routes on

demand leading to lower control overhead, but higher re-

sponse time, and (iii) hybrid (e.g., LQSR [32] and SrcRR [19])

where both approaches (i.e., proactive and reactive) are com-

bined to adjust the tradeoff between response time and con-

trol overhead.

Over the years, WMN routing protocols (e.g., AODV-ST,

SrcRR, B.A.T.M.A.N.) have evolved to target particular chal-

lenges or structures of WMNs. In particular, AODV-ST [50]

performs routing assuming a tree-based traffic flow, where a

gateway residing at the root of the tree (Internet access gate-

way) frequently requests routes to other nodes of the net-

work. SrcRR computes routes taking into account the effect

of interference on links quality. B.A.T.M.A.N. 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.

All these protocols can be integrated into our SDN archi-

tecture 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. In our experimental evalu-

ation, we rely on the WMN for the computation of routes be-

tween the local and the remote gateway (i.e., since the WMN

nodes do not support OpenFlow), while we use our algorithm

for gateway selection whenever traffic redirection is needed.

Our approach of coupling WMN routing protocols with SDN

is also employed by authors in [28,29,62].

Software-defined wireless networks. Recent work has been

leveraging on SDN for WMN management. Hasan et al. [36]

discuss the benefits of using SDN to decouple the opera-

tion and management of rural wireless networks from the

physical infrastructure. Our work is in the same direction,

but we focus on the management of crowd-shared WMN in

residential areas and investigate techniques for the efficient

utilization of the shared bandwidth. Authors in [28,29,62]

present OpenFlow-based architectures for WMNs with em-

phasis on mobility management, routing, and load balanc-

ing. Dimogerontakis et al. [30] propose a SDN extension to

WMN community networks for L2 experimentation. Other

applications of SDN to the wireless network domain include

OpenRoads [63] and OpenSDWN [59], which propose solu-

tions for mobility management using OpenFlow. These tech-

niques are complimentary to our work and can be integrated

into our SDN control plane to facilitate mobility management

in crowd-shared WMNs. Furthermore, authors in [52] inves-

tigate and evaluate the SDN control plane distribution for

WMNs with emphasis on master controller selection. This

can complement our work to provide higher resilience and

SDN controller availability.

Internet access sharing.FON [5], Kabel Deutschland (KD)

[17], 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

370 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

of shared hotspots. For example, BT and KD use Home

Hub and homespot, respectively, for Internet access sharing

with other subscribers. Furthermore, DT and BT have part-

nered with FON to provide Internet access through shared

home broadband connections. These efforts have also been

supported by several affordable, easy-to-install and open-

firmware wireless routers offered by manufacturers such as

FON [9], openMesh [15], Linksys [13], and Netgear [14]. Our

approach leverages on SDN to bring the following benefits to

Internet access sharing: (i) reduction of operational expenses

for home networks and ISPs, since the deployment, configu-

ration, and operation of crowd-shared networks can be out-

sourced to third-parties (i.e., virtual network operators), and

(ii) better coordination of network configurations for traffic

redirections, due to the global WMN view and the collection

of all monitoring and sharing policy information in a central-

ized controller.

In addition, Link-alike [40] promotes sharing by aggre-

gating the uplink bandwidth of the broadband connections

through WMN, such that home users have access to higher

upstream capacity. On the other hand, ARBOR [61] aggregates

broadband connection bandwidth (both uplink and down-

link) from multiple WiFi networks. We achieve the same goal

(i.e., high utilization of the shared bandwidth) via guest traf-

fic redirections through the WMN, enabling resource pooling

from all shared broadband connections.

7. Conclusions

In this paper, we investigated the benefits of extending

the coverage of any crowd-shared network (e.g., PAWS) by

connecting the home routers as a mesh. A crowd-shared

WMN can mitigate the fundamental problem of any crowd-

shared network, i.e., the presence of a single point of ac-

cess for each guest. We showed that the advance knowl-

edge of user sharing policies can reduce the number of guest

flow redirections (particularly for low and moderate network

loads) avoiding implications, such as packet reordering and

control communication overhead. This leads to a significantly

higher utilization of the shared bandwidth, as opposed to a

crowd-shared network with a single point of access where a

portion of the shared bandwidth is wasted. This, in turn, en-

ables a software-defined crowd-shared WMN to accommo-

date a substantially larger volume of guest traffic.

SDN brings significant benefits to crowd-shared net-

works, 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 redirec-

tion configurations in the WMN. In contrast, a decentralized

crowd-shared WMN management would require synchro-

nization primitives and could yield slow convergence, espe-

cially at the event of message loss. Our experimental results

show that our SDWMN control plane exhibits low per-flow

setup time and low control communication overhead.

Deploying a SDWMN in the real world raises challenges,

such as the controller placement, control plane distribu-

tion for robustness and load balancing, and SDN support

SDN in home routers. We envision hosting the SDN con-

troller as a virtual machine in micro-datacenters hosted by

large ISPs [35]. Micro-data centers are small DCs deployed

at POP and network aggregation points. By deploying the

controller in micro-DCs we ensure low communication de-

lay with the home routers and scalable processing power

to cope with high loads (scale up the controller computa-

tion capacity by using more servers). We can further leverage

on several frameworks to distribute the control load across

multiple servers, leading to faster response time and provid-

ing a fallback controller in case of failure [22,23,31,42,52]. In

terms of SDN support in home networks, there are already

OpenFlow-based wireless routers (e.g., HP [11], Meru [12]).

Furthermore, OpenFlow has been ported to OpenWRT [9,10],

allowing wireless routers running OpenWRT to be configured

using OpenFlow. This essentially facilitates the extension of

PAWS as a crowd-shared SDWMN, since PAWS already uses

wireless routers with OpenWRT-based firmware.

Content caching and the ever-growing penetration of on-

line social networks (OSN) create for more opportunities

in software-defined crowd-shared networks. More specifi-

cally, popular content can be placed in caches within home

networks, while the SDN controller can redirect guests to

caches, conserving bandwidth in shared broadband connec-

tions. Furthermore, information can be utilized from OSNs

in order to associate and authenticate guests to access points

that belong to their online friends, leveraging on the trust be-

tween two parties with an OSN relationship [26]. In future

work, we plan to investigate any potential benefits by cou-

pling content caching and OSNs with crowd-shared WMNs.

Acknowledgments

This work was partially supported by the EU CONFINE

Project (288535). We thank Amr Rizk for his help in modeling

the home router on/off periods.

References

[1] http://www.nrl.navy.mil/itd/ncs/products/mgen (accessed 18.03.15).

[2] CONFINE Project, http://confine-project.eu/ (accessed 18.03.15).[3] Community Lab, https://community-lab.net/ (accessed 18.03.15).

[4] http://www.awmn.net (accessed 18.03.15).[5] FON, http://corp.fon.com/ (accessed 18.03.15).

[6] Freifunk, http://freifunk.net/ (accessed 18.03.15).

[7] TFA Rice, http://tfa.rice.edu/ (accessed 18.03.15).[8] http://www.btwifi.com/help/about-bt-fon.jsp (accessed 18.03.15).

[9] http://wiki.openwrt.org/toh/start (accessed 18.03.15).[10] http://archive.openflow.org/wk/index.php/Pantou_:_OpenFlow_1.

0_for_OpenWRT (accessed 18.03.15).[11] http://h17007.www1.hp.com/docs/campus-lan/c04219571-hires.pdf

(accessed 18.03.15).

[12] http://www.merunetworks.com/collateral/white-papers/sdn-for-wifi-wp.pdf (accessed 18.03.15).

[13] http://www.linksys.com/gb/c/wireless-routers/ (accessed 18.03.15).[14] http://www.linksys.com/gb/c/wireless-routers/ (accessed 18.03.15).

[15] http://www.open-mesh.com/ (accessed 18.03.15).[16] https://www.telekom.com/media/consumer-products/180150 (acces-

sed 18.03.15).

[17] http://www.kabeldeutschland.de/wlan-hotspots/homespot-service.html (accessed 18.03.15).

[18] A. Abujoda, A. Sathiaseelan, A. Rizk, P. Papadimitriou, Software-definedcrowd-shared wireless mesh networks, in: IEEE International Work-

shop on Community Networks and Bottom-up-Broadband(CNBuB2014), Larnaca, Cyprus, 2014.

[19] D. Aguayo, J. Bicket, R. Morris, 2003, SrcRR: A High-Throughput Rout-

ing Protocol for 802.11 Mesh Networks. !http://pdos.csail.mit.edu/rtm/srcrr-draft.pdf?

[20] I. Akyildiz, X. Wang, W. Wang, Wireless mesh networks: a survey, Com-put. Netw. 47 (4) (2005) 445–487.

[21] E. Alotaibi, B. Mukherjee, A survey on routing algorithms for wirelessad-hoc and mesh networks, Comput. Netw. 56 (2012) 940–965.

A. Abujoda et al. / Computer Networks 93 (2015) 359–372 371

[22] Z. Bozakov, P. Papadimitriou, Autoslice: automated and scalable slicingfor software-defined networks, in: Proceedings of the 2012 ACM Con-

ference on CoNEXT Student Workshop, CoNEXT Student 12, ACM, 2012,p. 34.New York, NY, USA,

[23] Z. Bozakov, P. Papadimitriou, Towards a scalable software-defined net-work virtualization platform, in: Proceedings of the IEEE/IFIP SDNMO

2014, Krakow, Poland, 2014.

[24] B. Braem, et al., A Case for Research with and on community networks,ACM SIGCOMM CCR 43 (3) (2013).

[25] M.E.M. Campista, et al., Routing metrics and protocols for wirelessmesh networks, IEEE Netw. 22 (1) (2008) 6–12.

[26] Z. Cao, J. Fitschen, P. Papadimitriou, Social WiFi: hotspot sharing withonline friends, in: Proceedings of the IEEE PIMRC, Hong Kong, China,

2015.[27] T. Clausen, P. Jacquet, 2003, Optimized Link State Routing Protocol

(OLSR),RFC 3626 (Experimental), IETF.

[28] P. Dely, A. Kassler, N. Bayer, Openflow for wireless mesh networks, in:Proceedings of the IEEE ICCCN, Maui, USA, 2011.

[29] A. Detti, C. Pisa, S. Salsano, N. Blefari-Melazzi, Wireless Mesh softwaredefined networks (wmSDN), in: Proceedings of the CNBuB Workshop

at IEEE WiMob, France, Lyon, 7, 2013.[30] E. Dimogerontakis, I. Vilata, L. Navarro, Software defined networking

for community network testbeds, in: Proceedings of the IEEE WiMob,

2013.[31] A. Dixit, F. Hao, S. Mukherjee, T.V. Lakshman, R. Kompella, Towards

an elastic distributed SDN controller, in: Proceedings of the ACM SIG-COMM Workshop on Hot Topics in Software Defined Networking,

HotSDN, 2013.[32] R. Draves, J. Padhye, B. Zill, Routing in multi-radio, multi-hop wireless

mesh networks, in: Proceedings of the ACM International Conference

on Mobile Computing and Networking (MobiCom), 2004, pp. 114–128.[33] S. Even, A. Itai, A. Shamir, On the complexity of timetable and multi-

commodity flow problems, SIAM J. Comput. 5 (1976) 691–703.[34] J. Fildes, 2010, UN reveals global disparity in broadband access, BBC,

http://www.bbc.co.uk/news/technology-11162656.[35] B. Frank, I. Poese, G. Smaragdakis, A. Feldmann, B. Maggs, S. Uhlig,

V. Aggarwal, F. Schneider, Collaboration opportunities for content de-

livery and network infrastructures, in: H. Haddadi, O. Bonaventure(Eds.), ACM SIGCOMM ebook on Recent Advances in Networking, 2013,

pp. 305–377.[36] S. Hasan, Y. Ben-David, C. Scott, E. Brewer, S. Shenker, Enhancing rural

connectivity with software defined networks, in: Proceedings of theAnnual Symposium on Computing for Development ACM DEV, New

York, NY, USA, 2013.

[37] N. Hu, P. Steenkiste, Evaluation and characterization of available band-width probing techniques, in: Proceedings of the IEEE JSAC, 2003.

[38] D. Johnson, N. Ntlatlapa, C. Aichele, Simple pragmatic approach to meshrouting using batman, in: Proceedings of the 2nd IFIP International

Symposium on Wireless Communications and Information Technologyin Developing Countries, CSIR, Pretoria, South Africa, 6–7 October 2008,

2008.

[39] M. Jain, C. Dovrolis, Pathload: A measurement tool for end-to-end avail-able bandwidth, in: Proceedings of the Passive and Active Measure-

ments (PAM) Workshop, 2002.[40] S. Jakubczak, D. Andersen, M. Kaminsky, K. Papagiannaki, S. Seshan,

Link-alike: using wireless to share network resources in a neighbor-hood, in: MC2R, 2008.

[41] E. Kohler, R. Morris, B. Chen, J. Jahnotti, M.F. Kasshoek, The Click Mod-ular Router, ACM Trans. Comput. Syst. 18 (3) (2000).

[42] T. Koponen, M. Casado, N. Gude, J. Stribling, L. Poutievski, M. Zhu, R. Ra-

manathan, Y. Iwata, H. Inouye, T. Hama, S. Shenker, Onix: a distributedcontrol platform for,large-scale production networks, in: Operating

Systems, Design and Implementation, USENIX association, 2010.[43] N. McKeown, et al., OpenFlow: enabling innovation in campus net-

works, ACM SIGCOMM CCR 38 (2) (2008).[44] B. Melander, M. Bjorkman, P. Gunningberg, A new end-to-end prob-

ing and analysis method for estimating bandwidth bottlenecks, in: Pro-

ceedings of the Global Internet Symposium, 2000.[45] B. Milic, M. Malek, NPART - Node Placement Algorithm for Realistic

Topologies in Wireless Multihop Network Simulation, in: Proceedingsof the 2nd International Conference on Simulation Tools and Tech-

niques, 2009.[46] S. Murthy, J.J. Garcia-Luna-Aveces, An efficient routing protocol for

wireless networks, ACM/Baltzer Journal on Mobile Networks and Ap-

plications, Spec. Issue Routing Mob. Commun. Netw. 1 (2) (1996) 183–197.

[47] C. Perkins, E. Belding-Royer, S. Das, Ad hoc on-demand distance vector(AODV) routing, in: RFC 3561 (Experimental)„ IETF, 2003.

[48] C.E. Perkins, P. Bhagwat, Highly dynamic destination sequenceddistance-vector routing (DSDV) for mobile computers, in: Proceedings

of the Annual ACM Special Interest Group on Data Communications(SIGCOMM94), 1994, pp. 234–244.

[49] B. Pfaff, J. Pettit, T. Koponen, K. Amidon, M. Casado, S. Shenker, Ex-tending networking into the virtualization layer, in: ACM HotNets, New

York, USA, ACM, 2009.

[50] K.N. Ramachandran, M.M. Buddhikot, G. Chandranmenon, S. Miller,E.M. Belding-Royer, K.C. Almeroth, On the design and implementation

of infrastructure mesh networks, in: Proceedings of the IEEE Workshopon Wireless Mesh Networks (WiMesh), IEEE, 2005.

[51] J. Saldana, et al., 2014, Community Networks: Definition and taxonomy,Internet Draft.

[52] S. Salsano, G. Siracusano, A. Detti, C. Pisa, P.L. Ventre, N. Blefari-Melazzi,Controller selection in a Wireless Mesh SDN under network partition-

ing and merging scenarios, submitted paper, available at http://arxiv.

org/abs/1406.2470.[53] A. Sathiaseelan, et al., Public access WiFi service (PAWS), in: Digital

Economy All Hands Meeting, Digital Futures, Aberdeen, 2012.[54] A. Sathiaseelan, C. Rotsos, C.S. Sriram, D. Trossen, P. Papadimitriou,

J. Crowcroft, Virtual public networks, in: Proceedings of the IEEE Eu-ropean Workshop on Software Defined Networks EWSDN, Berlin, Ger-

many, 2013.

[55] A. Sathiaseelan, J. Crowcroft, Internet on the move: challenges and so-lutions, ACM SIGCOMM CCR 43 (1) (2013).

[56] A. Sathiaseelan, T. Radzik, Reorder detecting TCP, in: Proceedings of theIEEE International Conference on HSNMC, Estoril, Portugal, 2003.

[57] A. Sathiaseelan, T. Radzik, Reorder notifying TCP (RN-TCP) with ex-plicit packet drop notification (EPDN), Int. J. Commun. Syst. 19 (6)

(2005).Wiley

[58] A. Sathiaseelan, J. Crowcroft, LCD-Net: lowest cost denominator net-working, ACM SIGCOMM CCR 43 (2) (2013).

[59] J. Schulz-Zander, C. Mayer, B. Ciobotaru, S. Schmid, A. Feldmann,OpenSDWN: programmatic control over home and enterprise WiFi, in:

Proceedings of the ACM SIGCOMM SOSR, Santa Clara, CA, USA, 2015.[60] L. Townsend, A. Sathiaseelan, G. Fairhurst, C. Wallace, Enhanced broad-

band access as a solution to the social and economic problems of the

rural digital divide, J. Local Econ. 28 (6) (2013).[61] X. Xing, S. Mishra, X. Liu, ARBOR: hang together rather than hang sep-

arately in 802.11 wifi networks, in: Proceedings of the IEEE INFOCOM,2010, 2010, pp. 1–9.IEEE

[62] F. Yang, V. Gondi, J.O. Hallstrom, K. Wang, G. Eidson, OpenFlow-basedload balancing for wireless mesh infrastructure, in: Proceedings of the

IEEE Consumer Communications and Networking Conference, CCNC,

Las Vegas, 2014.[63] K. Yap, et al., OpenRoads: empowering research in mobile networks, in:

Proceedings of the ACM SIGCOMM, Barcelona, Spain, 2009.

Ahmed Abujoda graduated in electrical engi-neering from the university of Sana’a, Yemen, in

2004. He also obtained a master of science de-gree in Mechatronics engineering from University

of Siegen, Germany, in 2009. Since 2010, he hasbeen working towards the Ph.D. degree in elec-

trical and computer engineering at the Institute

for Communications Technology at the LeibnizUniversitt Hannover, Germany. His research inter-

ests include software-defined networking, net-work function virtualization and network man-

agement. Ahmed Abujoda is a student member ofIEEE.

David Dietrich graduated in electrical engineer-

ing from the Universitt Kassel, Germany, in 2006.During the years 2006 until 2009 he was a soft-

ware engineer for real-time simulation of auto-motive communications networks. Since 2010, he

has been working towards the Ph.D. degree in

electrical and computer engineering at the Insti-tute for Communications Technology at the Leib-

niz Universitt Hannover, Germany. He also partic-ipated in several research projects, e.g., the EU-

funded T-NOVA project and the national projectG-Lab. His research interests include network vir-

tualization and network management with focus

on algorithms and optimization. David Dietrich is a student member of IEEEand ACM.

372 A. Abujoda et al. / Computer Networks 93 (2015) 359–372

Panagiotis Papadimitriou is an Assistant Pro-

fessor at the Communications Technology Insti-tute of Leibniz Universitt Hannover. He received

a Ph.D. in Electrical and Computer Engineering

from Democritus University of Thrace, Greece, in2008. From 2008 to 2010, Panagiotis was a Post-

Doctoral Researcher at the Computing Depart-ment of Lancaster University, UK, and at Telecom

SudParis, France. Panagiotis received the runner-up Poster Award at ACM SIGCOMM 2009 and the

Best Paper Award at IFIP WWIC 2012. The re-

search activities of his group currently span next-generation Internet architectures, network func-

tion virtualization, software-defined networks, and cloud networking.

Arjuna Sathiaseelan is a Senior Research Asso-

ciate at the Computer Laboratory, University ofCambridge. He leads the Networking for Develop-

ment (N4D Lab). The research group conducts re-

search on novel Internet architectures for improv-ing and reducing the cost of Internet access. He is

the Chair of IRTF Global Access to the Internet forAll (GAIA) research group and a member of the

Internet Research Steering Group (IRSG). He hasa Ph.D. in Networking from Kings College London

(2005), M.Sc. in Computing and Internet Systems

from Kings College London (2001) and Bachelorsin Computer Science and Engineering from NIT,

Trichy, India (2000).