Ad Hoc Networks 9 (2011) 1256–1269

Contents lists available at ScienceDirect

Ad Hoc Networks
journal homepage: www.elsevier.com/locate/adhoc

Joint routing and scheduling optimization in arbitrary ad hoc

networks: Comparison of cooperative and hop-by-hop forwarding
Antonio Capone a,⇑, Stefano Gualandi a, Di Yuan b
a
Dipartimento di Elettronica e Informatica, Politecnico di Milano, Viale Ponzio 24/A, 20133 Milan, Italy
b
Department of Science and Technology, Linköping Institute of Technology, SE-601 74 Norrköping, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: Cooperation schemes form a key aspect of infrastructure-less wireless networks that allow
Received 10 June 2010 nodes that cannot directly communicate to exchange information through the help of
Received in revised form 30 December 2010 intermediate nodes. The most widely adopted approach is based on hop-by-hop forward-
Accepted 6 February 2011
ing at the network layer along a path to destination. Cooperative relaying brings coopera-
Available online 22 February 2011
tion to the physical layer in order to fully exploit wireless resources. The concept exploits
channel diversity by using multiple radio units to transmit the same message. The under-
Keywords:
lying fundamentals of cooperative relaying have been quite well-studied from a transmis-
Cooperative networking
Routing and scheduling
sion efficiency point of view, in particular with a single pair of source and destination.
Column generation Results of its performance gain in a multi-hop networking context with multiple sources
Cooperative relaying and destinations are, however, less available. In this paper, we provide an optimization
Optimization approach to assess the performance gain of cooperative relaying vis-a-vis conventional
multi-hop forwarding under arbitrary network topology. The approach joint optimizes
packet routing and transmission scheduling, and generalizes classical optimization
schemes for non-cooperative networks. We provide numerical results demonstrating that
the gain of cooperative relaying in networking scenarios is in general rather small and
decreases when network connectivity and the number of traffic flows increase, due to
interference and resource reuse limitations. In addition to quantifying the performance
gain, our approach leads to a new framework for optimizing routing and scheduling in
cooperative networks under a generalized Spacial Time Division Multiple Access (STDMA)
scheme.
Ó 2011 Elsevier B.V. All rights reserved.

1. Introduction through intermediate nodes that relay messages towards

their destinations.
Infrastructure-less wireless networks have received The first and most common cooperation mechanism is
much attention from the research community and gener- based on hop-by-hop forwarding along a path composed
ated a plethora of applications (including ad hoc, sensor, by a sequence of nodes from the source to the destination.
and mesh networks). In most of these application scenar- At each hop, a single node forwards an information packet
ios, direct communication is not possible for all node pairs to the next node in the path which is usually dynamically
due to the transmission range limitations, and thus cooper- selected through a routing protocol [1]. This conventional
ative approaches are adopted. The goal is to allow nodes multi-hop forwarding scheme has been adopted in almost
that cannot directly communicate to exchange information all practical implementations of ad hoc, sensor and wire-
less mesh networks, and analyzed by a huge amount of
⇑ Corresponding author. Tel.: +39 02 2399 3449; fax: +39 02 2399 3413. literature.
E-mail addresses: [email protected] (A. Capone), gualandi@elet. Recently, in order to overcome the capacity limits of
polimi.it (S. Gualandi), [email protected] (D. Yuan). conventional multi-hop forwarding, a new approach

1570-8705/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.adhoc.2011.02.002
 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1257

exploiting cooperation also at the transmission layer has

been considered [2,3]. The idea, that dates back to the work
of van der Meulen [4] and Cover [5], is to use multiple
relays whose transmissions are combined together at the
receiver. Exploiting the broadcast nature of the radio
channel, transmissions can be received and then simulta-
neously forwarded by groups of nodes, each acting as a
virtual array of antennas [6,7]. Relay nodes can decode-
and-forward or just amplify-and-forward information
packets [8,9].
Similarly to multiple antenna systems, with cooperative
relaying there are some remarkable advantages in terms of
channel capacity [10–12]. However, whereas the perfor- (a) (b)
mance gain of cooperative relaying is easy to understand
in single source–destination scenarios, the case of multiple Fig. 1. Cooperation and resource reuse trade-off: (a) all nodes cooperate
sources and destinations is more complex to analyze for a single transmission, (b) partial cooperation with parallel
[21,22] and requires the design of cooperation protocols transmissions.

for multiple access, channel state information, and routing

[23–25]. In general, it is expected that the signaling over-
head for cooperation information is higher than traditional  We account for interference with the physical model
hop-by-hop forwarding. For some specific scenarios, it has that considers a minimum threshold on the signal-to-
been shown that protocols for cooperation may require interference and noise ratio (SINR) [26]. With coopera-
twice the bandwidth in comparison to non-cooperative tion relaying, we use the sum of the contributing signals
schemes [8]. However, protocol overhead is not the only in the overall SINR. This allows to evaluate accurately
limiting factor of the performance of cooperative relaying. the effect of resource reuse without the limitations of
In this paper we aim to show that, even neglecting the im- the protocol interference model.
pact of increased protocol overhead, from a network capac-  We capture the channel quality improvement of coop-
ity perspective with multiple sources and destinations, the erative relaying in terms of increased transmission
potential gain of cooperative relaying may be quite small. range. The model can be extended to a discrete set of
This is due to the interference generated by simultaneous multiple transmission rates and SINR thresholds at the
transmissions and the limitations caused by half-duplex cost of a higher complexity.
operation of wireless devices.  Since both packet routing and transmission scheduling
Indeed, from a network point of view, in addition to greatly impact the overall network capacity, we jointly
radio channel capacity, there is another important element optimize them considering a Spacial Time Division Mul-
that determines the overall efficiency: the reuse level of tiple Access (STDMA) approach in order to make the
radio resources in the network. Parallel transmissions of comparison fair. This also allows us to neglect the effect
nodes are possible if they are sufficiently apart and the of protocol overheads in the analysis.
interference at their receivers is low enough to allow for  Our model works for arbitrary network topologies and
correct decoding of information packets. We argue that, optimizes the path selection and the transmission
even if cooperative relaying can increase the radio channel scheme at each hop with or without cooperation. There-
capacity, or equivalently the transmission range, it may fore, for the cooperation case, the optimization includes
negatively affect the reuse level. Transmissions from mul- the decision of optimal relay selection.
tiple relays generate interference that adds to the noise,  We assume half-duplex operation of wireless devices
thus limiting the number of parallel transmissions of dif- that, therefore, cannot transmit and receive at the same
ferent packets. Moreover, since nodes are usually equipped time.
with a single transceiver, at a given time they can be in-
volved in only one cooperation group for transmitting or It is easy to see that most of the above assumptions de-
receiving a packet, further limiting resource reuse. In the fine a comparison scenario that is favorable to cooperative
extreme case, all nodes cooperate to relay the packet of a relaying, and the conventional multi-hop forwarding
source–destination pair, and no parallel transmission can scheme is just a special case of our cooperative relaying
occur in the network (Fig. 1). framework. This allows us to measure the relative gain in
In this paper, we provide an optimization approach to network efficiency selecting the best level of cooperation.
compare the performance of cooperative relaying with that The cooperation gain we calculate can be considered an
of conventional multi-hop forwarding. To this end, we upper limit that may not be reachable in practice due to
have adopted the following modeling approach: the overhead required by cooperation. We show, however,
that even for this upper limit, the cooperation gain may be
 We consider a decode-and-forward mechanism. The quite small in some network settings where multiple
model can be extended to the case of amplify-and- connections must be served in parallel and the interference
forward, although it would lead to a non-linear and level is high.
discrete optimization problem that is beyond the scope It is worth pointing out that the optimization frame-
of the current paper. work we present does not define a new cooperation
1258 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269

algorithm, and hence the research theme differs from the In [29], the gain of cooperative relaying in large net-
scope of many practical cooperation algorithms that have works with multiple sources and destinations and ampli-
been proposed in the literature. The proposed approach ex- fy-and-forward relays is evaluated, and asymptotic
tends and generalizes classical optimization schemes for results are provided. Interference is modeled using a con-
routing and scheduling in non-cooperative networks [30– flict graph according to the protocol model. In our work,
33]. Therefore, it also has a practical value in those network we address interference with the more accurate physical
scenarios where propagation and traffic conditions are model based on SINR, and propose a different approach
quite stable and a centralized scheme can be used to opti- for evaluating cooperative relaying and optimizing re-
mize radio resources, like for example in wireless mesh source management, in order to provide a method for
networks. To the best of our knowledge, this is the first ap- defining optimal scheduling, relay selection and routing
proach that provides a framework for optimization of radio schemes in arbitrary-topology networks.
resources (routing, scheduling, and relay selection) in
cooperative relaying and that allows a vis-a-vis compari-
3. System model
son with conventional multi-hop forwarding.
The paper is organized as follows. Section 2 is devoted
3.1. Model overview
to the review of related work. In Section 3 we present
the system model and the related assumptions. In Section
The approach we use to define a mathematical model of
4 we propose an integer programming formulation of the
the system and to formulate the resource optimization
cooperation optimization problem, and in Section 5 we
problem uses two types of graphs. Specifically, we define
present a solution approach based on column generation.
a Physical Topology Graph and a set of Cooperation Graphs
Computation results on some illustrative examples and
with different levels j of cooperation.
on a larger set of randomly generated networks are pro-
The physical topology graph defines the set of possible
vided in Section 6. Finally, Section 7 concludes the paper.
direct communications between wireless devices (graph
nodes) without any cooperative relaying. In the graph, a
2. Related work
(direct) link exists if and only if the signal-to-noise ratio
(SNR) at the receiving node is above a threshold, and the
There is a quite large amount of literature on the chan-
interference constraint is defined according to the classical
nel capacity of various cooperative relaying schemes. Re-
physical model [26].
search in this area has taken into consideration several
The cooperation graph defines the set of possible direct
aspects including distributed coding schemes [13–15],
communications between wireless devices in case of coop-
retransmission mechanisms [16], and power allocation
erative relaying. In this graph, a node corresponds to a set
methods [17,18]. In addition to capacity analysis, it has
of cooperating devices, and a link defines the possible di-
been shown that cooperative relaying can provide some
rect transmission from a set of transmitting devices to a
spacial diversity gain due to the fading effect of the radio
set of receiving devices. The level of cooperation is defined
channel caused by obstacles and multi-path propagation
by the maximum cardinality of the set of cooperating de-
[8,19], even though there is an obvious trade-off between
vices that may form a node in the cooperation graph. The
capacity and diversity gains [20].
existence of a link in the cooperation graph depends on
On the other hand, network-level evaluation of the per-
the SNR at all the receiving devices, where the signals of
formance gain of cooperative relaying in wireless multi-
all transmitting devices are summed up. The interference
hop network has received so far little attention in the liter-
constraint is defined by extending the physical model
ature. It amounts to measuring the overall efficiency in
and considering the SINR at all receiving nodes, where
using radio resources when serving multiple traffic flows
the signal contributions from all transmitting nodes are
between different source–destination pairs in presence of
put together. In the following we present the system model
interference. In fact, due to the difficulty of modeling inter-
in a mathematically formal way.
ference, network capacity evaluation is not an easy task
even with the conventional multi-hop forwarding scheme,
for which mainly asymptotic results exist [26]. 3.2. Physical topology graph
A first work in [27] considers cooperative relaying in a
network with two sources and two destinations, and eval- The physical topology is represented by a directed
uates the channel capacity. Differently from [27], our study graph G ¼ ðN ; LÞ, where N and L denote the sets of nodes
is characterized by the consideration of arbitrary ad hoc and communication links, respectively. Denote by Pv the
network topologies and the focus on the evaluation of net- transmission power of v 2 N ; g v w the radio propagation
work capacity. gain between v and w, and g the effect of thermal noise.
Modeling the interference from a statistical perspective, A link (v, w) exists if and only if its SNR is above a threshold
the authors of [28] consider ad hoc network scenarios with c:
multi-user interference and analyze the behavior of the Pv g v w
amplify-and-forward scheme. We consider instead the SNRv w ¼ P c: ð1Þ
g
decode-and-forward scheme and evaluate the network
efficiency in using radio resources when simultaneous With time division multiple access, radio transmissions are
transmissions can be coordinated and interference- organized using time slots, each of which may accommo-
controlled through an optimal scheduling scheme. date one or several parallel transmissions. Transmission
 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1259

between two nodes generates interference at the other Henceforth, we simplify the notation of the j-coopera-
receivers. Suppose link (v, w) is used for a transmission tion graph to G, when there is no ambiguity. For the sake
from v to w, and K is the set of transmitting nodes in addi- of clarity, we use v and w to denote nodes in the original
tion to v in the same time slot. The transmission on (v, w) is graph G, and i and j nodes in the cooperation graph G. Nodes
considered successful if and only if the signal-to-interfer- and links in G are also referred to as super-nodes and super-
ence-and-noise ratio (SINR) is at least c, i.e., links. Let C(i) denote the set of nodes in N forming super-
Pv gv w node i in V, and K(i, j) the set of links forming super-link
SINRv w ¼ P P c: ð2Þ (i, j) 2 A: K(i, j) = {(v, w)jv 2 C(i),w 2 C(j)}. Note that the size
u2K P u g uw þg
of G grows exponentially in j. When j = 1, the j-coopera-
tion graph G reduces to the original topology G. The concepts
3.3. Performance metric and problem definition of super-node and super-link are illustrated in Fig. 2. In this
example, each of the two super-nodes i and j contains two
A widely used performance metric in scheduling trans- nodes in N , and the super-link (i, j) represents transmissions
missions is the number of time slots required to deliver from all nodes in C(i) to all nodes in C(j).
traffic demand. The metric is directly related to the overall In Fig. 2, the four transmissions do not necessarily cor-
efficiency in using radio resources. In comparing coopera- respond to links in the original graph G. This is because the
tive relaying with conventional forwarding, this metric SNR condition takes a new form in the cooperation graph
captures the efficiency of resource reuse, i.e., the minimi- G. For a super-link (i, j) 2 A to exist, the following SNR con-
zation of the number of slots through parallel transmis- dition applies to all the receivers of the super-link, i.e., for
sions, as well as the effect of range extension by all w 2 C(j):
cooperative relaying, which potentially reduces the re- P
quired number of transmissions. v 2CðiÞ P v g v w
To this end, we consider a demand set D; each element SNRiw ¼ P c: ð3Þ
g
d 2 D is a packet associated with a tuple hod, tdi, where od
are td are the source and destination of the packet, respec- The numerator in (3) models the fact that the nodes in C(i)
tively. Without cooperative relaying, each transmission in- are transmitting the same packet and hence all contribut-
volves exactly one sender and one receiver, and takes place ing to improving SNR. For this reason, a super-link can be
over one link in L. With cooperation in transmission, mul- established, as a result of cooperation, even if some or pos-
tiple nodes may simultaneously transmit or receive the sibly none of the transmissions of this super-link is part of
same packet, and a receiver can combine the signals from the link set in the original topology.
multiple senders, which may or may not have links to When several super-nodes are transmitting in the same
the receiver in the physical topology G. time slot, interference must be accounted for. Suppose in
addition to i, a set of super-nodes X are transmitting in
Definition 1. A wireless network G has a j-level of the same time slot. The SINR condition for super-link (i, j)
cooperation, if, during the same time slot, at most j is that the following inequality holds for all w 2 C(j):
nodes are allowed to transmit the same packet to a set of P
one or more receiving nodes, and each receiver can Pv g v w
SINRiw ¼ P Pv 2CðiÞ P c: ð4Þ
combine the signals from the j transmitters. l2X u2CðlÞ P u g uw þg
Our comparative study of network performance seeks
the answer to the following question: What are the mini- In (4), all nodes composing the super-nodes in X generate
mum numbers of time slots required to deliver the given interference to super-link (i, j). Note that a node u 2 N may
set of packets with and without cooperation, respectively? appear at most once in the denominator, because the
This leads to an optimization problem in routing and super-nodes transmitting in any time slot all must have
scheduling the packet set D in network G under a j-level mutually disjoint sets of nodes in the original graph.
of cooperation, where a non-cooperative network corre-
sponds to j = 1. 3.5. Classes of super-links

3.4. Communication graph with cooperation The super-nodes in G vary in the cardinality of the asso-
ciated subsets of nodes in the original graph. Based on this
For j > 1, graph G is not sufficient for modeling cooper- cardinality, we define four classes of super-links.
ative transmission. We develop a graph concept, which we
refer to as the j-cooperation graph, to generalize the origi-
nal topology G.

Definition 2. For graph G, the j -Cooperation Graph

GG;j ¼ ðV; AÞ, is the auxiliary graph representing all possi-
ble transmissions in G permitted by a j-level of coopera-
tion. A node i 2 V represents a non-empty subset of N
having cardinality up to j, and a link (i, j) 2 A represents
simultaneous transmission of one packet from all nodes in
the i’s node set in N to j’s node set in N . Fig. 2. An illustration of super-nodes and super-link.
1260 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269

4. Many-to-many: This class of links in G represents

transmissions of the same packet between multiple
transmitters and receivers in the original graph G. A
super-link (i, j) is a many-to-many link if and only if
jC(i)j > 1 and jC(j)j > 1. The super-link shown in
Fig. 2 is of this class. Similar to many-to-one super-
links, a many-to-many super-link (i, j) can be created
by cooperation, hence it may have transmissions
between one or multiple pairs of nodes v 2 C(i) and
w 2 C(j) for which ðv ; wÞ R L. Many-to-many super-
links are also called broad-cooperating links or multi-
casting links.
Fig. 3. Topology of a very simple wireless network.

Example 1. Fig. 3 shows the physical topology of a wire-

1. One-to-one. These are the links in the original physical less network with five nodes. Fig. 4 shows the super-nodes
topology, i.e., links in L. Super-link (i, j) is of this class if of the corresponding 2-cooperation network. For clarity,
it satisfies the condition: C(i) = {v}, C(j) = {w}, and only a few of the super-links are drawn. The dot-dashed
ðv ; wÞ 2 L. links are of class one-to-one; these are the links incident
2. One-to-many. A super-link of this class corresponds to to node 2 in the original topology. The solid super-links
a group of links in L originating from the same node in are broadcasting links from node 3 to super-nodes {1, 2},
N . Thus (i, j) 2 A is a one-to-many super-link if {1, 5}, and {2, 5}, respectively. Each of the three dotted
C(i) = {v}, jC(j)j > 1, and ðv ; wÞ 2 L for all w 2 C(j). The super-links is a cooperating link, representing simulta-
one-to-many links are also referred to as broadcast links. neous transmissions of a packet from super-node {2, 3} to
A special subclass of one-to-many links is the so called a single receiver. Note that transmissions on (2, 1), (2, 5),
buffering links, in which a node behaves as if it is trans- and (3, 4) do not correspond to any physical link in Fig. 3.
mitting also to itself. Finally, the dashed super-links are multicasting links, each
3. Many-to-one. A super-link (i, j) 2 A is of class many-to- of which carries transmissions involving two transmitters
one, if jC(i)j > 1 and CðjÞ ¼ fwg; w 2 N n CðiÞ. This and two receivers; some of these transmissions occur
super-link represents simultaneous transmissions of the between nodes that do not have links in Fig. 3. The distance
same packet from all nodes in C(i) to the single receiver in hops between a source–destination pair may benefit
w in C(j). A many-to-one link does not necessarily consist from cooperating and multicasting links. For example, the
in a group of links in the original graph, i.e., C(i) may con- shortest path from node 4 to node 1 has three hops if no
tain node v for which ðv ; wÞ R L, provided that the SNR at cooperation is allowed. The shortest path distance in the
w satisfies (3) by cooperation. Links in the many-to-one cooperation graph becomes two hops, and one of such
class are also referred to as cooperating links. paths is formed by 4, {2, 5}, and 1.

Fig. 4. Super-nodes and some super-links of the 2-cooperation graph for the network of Fig. 3.
 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1261

4. An integer programming formulation routing from the sources to the destinations, and incorpo-
rate the decision of relay selection in the cooperative
Routing and scheduling of packets without cooperation, scheme. Constraints (7) link the flow variables b to the con-
with the objective of minimizing the total number of time figuration variables k: for each super-link (i, j) 2 A, suffi-
slots, has been studied quite extensively in the literature ciently many configurations containing (i, j) must be
(see Section 1). From a computational complexity stand- chosen to accommodate the total flow on (i, j).
point, this optimization problem is NP-hard, in particular
because even if optimization is restricted to scheduling 5. A column generation approach
only (i.e., routing is fixed), the problem generalizes graph
coloring [30]. 5.1. The framework
It has been shown that integer programming formula-
tions based on identifying feasible configurations admits, The number of k-variables in P equals jSj, which is
within a reasonable amount of computing time, an optimal exponential in the size of the cooperation graph G. Thus a
or very-close-to optimal routing and scheduling solution solution algorithm requiring the generation of all possible
for networks of up to moderate size [32,33]. feasible configurations a priori is not computationally fea-
The key element in these formulations is the definition sible. Exploiting the structure of P, an effective solution
of a feasible configuration that refers to a group of trans- strategy is to solve its linear programming (LP) relaxation
missions that may share a time slot (simultaneous trans- by means of a column generation method, which provides
missions) without violating the SINR constraint at any a lower bound on the optimal number of time slots, fol-
receiver. Each feasible configuration is associated with a lowed by computing the optimal or a near-optimal integer
0–1 decision variable, and the problem formulation defines solution (see, e.g., [34] for a textbook on column genera-
which configurations to activate in different time slots in tion). By keeping a small subset S 0  S and expanding the
order to minimize their total number. subset systematically, the column generation method
Utilizing the cooperation graph, we are able to general- decomposes the LP version of P into a master problem
ize this integer programming concept to networks with and a subproblem. The former is used to find the optimal
cooperation. For cooperation graph G, a feasible configura- solution when the optimization is restricted to the ele-
tion of a time slot is a group of super-links, such that the ments in S 0 . The latter corresponds to a separation problem
SINR condition (4) is satisfied for all of them under simul- for the dual LP for checking optimality, that is, to either
taneous transmissions. Thus constructing a scheduling find additional elements that are of interest to be added
solution amounts to selecting a set of feasible configura- to S 0 , or conclude that the current solution of the master
tions, and the cardinality of the set equals the total number problem is optimal not only for S 0 but also for S (i.e., all
of time slots. variables ks for s 2 S n S 0 are zeros at optimum).
Let S denote the collection of all feasible configurations, The master problem looks very similar to P. To save
and Sij # S the set of configurations containing super-link space, we do not present it in mathematical form. The mas-
d
(i, j) 2 A. Let bi be a parameter that is equal to 1 if node ter problem differs from P in the following two aspects.
i 2 N \ V is the source of the packet d, 1 if it is the desti-
nation of d, and 0 otherwise. We define two sets of optimi- 1. Set S is replaced by a subset S 0 , of which jS 0 j  jSj.
zation variables. 2. The integrality requirements on the b- and k-variables
are relaxed.
ks = the number of time slots allocated to configuration
s 2 S, As long as the resulting master problem contains at
bdij ¼ f 1 if super  link ði; jÞ is used to route packet least one feasible solution, any S 0  S can serve as the
d from i to j; 0 otherwise: starting point. That is, the subset S 0 must be such that con-
straints (7) can be satisfied. The simplest choice consists of
The problem of routing and scheduling the demand set setting S 0 ¼ A, that is for each super-link (i, j) in A we have a
D with a minimum number of time slots can be formulated configuration where (i, j) is the only active link.
as the following integer linear programming model.
X 5.2. The pricing problem
½P min ks ð5Þ
s2S
X X d After solving the master problem, checking the LP opti-
s:t: bdij  bdji ¼ bi ; i 2 V; d 2 D; ð6Þ
mality condition amounts to examining whether or not
ði;jÞ2A ðj;iÞ2A
X X any ks ; s 2 S n S 0 , has a negative reduced cost. Denoting
ks P bdij ; ði; jÞ 2 A; ð7Þ the dual variables of (7) by p  ij ; ði; jÞ 2 A, the reduced cost
s2Sij d2D P
of ks equals 1  ði;jÞ2A:s2Sij p
 ij . The subproblem, also called
bdij 2 f0; 1g; ði; jÞ 2 A; d 2 D; ð8Þ the pricing problem, takes the following form,
X
ks 2 Zþ ; s 2 S: ð9Þ min ð1  p ij Þ: ð10Þ
s2SnS 0
ði;jÞ2A:s2Sij
The objective is to minimize the total number of configura-
tions (i.e., time slots). Constraints (6) are the classical flow The minimization in (10), in its turn, can be formulated by
balance equations for each packet d 2 D. They define the an integer linear model; thus we can solve (10) without
1262 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269

8
explicitly having the set S. A solution of this integer model < 1 if transmission of node
> v 2N gives
corresponds to a feasible configuration, that is, a set of q0v w ¼ interference at w 2 N ;
super-links that can be active in the same time slot. Below >
:
0 otherwise
we define the set of binary variables used by the model,
and the model itself. 
 1 if node v2N is transmitting;
1 if super  link ði; jÞ is active; xv ¼
zij ¼ 0 otherwise:
0 otherwise:

X 1 if node w 2 N is receiving;
½R1 min  p ij zij ð11Þ yw ¼
0 otherwise:
ði;jÞ2A
X
X X ½R2 min  p ij zij
s:t: zij þ zji 6 1; v 2 N; ð12Þ ði;jÞ2A
ði;jÞ2A: ðj;iÞ2A:
v 2CðiÞ v 2CðiÞ
s:t: ð12Þ;
P
v 2CðiÞ P v g v w X
P P P czij ;
ðk;lÞ2A: u2CðkÞ P u g uw zkl þg zij ¼ qv w ; v 2 N ; w 2 N ; v –w;
k–i;l–j ði; jÞ 2 A :
ði; jÞ 2 A; w 2 CðjÞ; ð13Þ ð16Þ
v 2 CðiÞ;
w 2 CðjÞ
zij 2 f0; 1g; ði; jÞ 2 A: ð14Þ
The objective is to find a configuration of minimum re- xv ¼ qv w þ q0v w ; v 2 N ; w 2 N ; v –w; ð17Þ
duced cost. Constraints (12) ensure that the z-variables
satisfy two conditions. First, for any node v in the original qv w 6 yw ; v 2 N ; w 2 N ; v –w; ð18Þ
topology G, at most one super-link, of which the starting
P
super-node contains v, can be active. In effect, v is allowed P g q
to transmit at most one packet in a time slot with or with- P v 2N v v w0 v w P cyw ; w 2 N ; ð19Þ
u2N P u g uw quw þ g
out cooperation. Second, if node v is transmitting, it cannot
be a receiving node of any active super-link at the same
zij 2 f0; 1g; ði; jÞ 2 A; ð20Þ
time. The SINR conditions are formulated by (13). The con-
straint has effect only if zij = 1. In this case, the inequality
states that the SINR, in which the interference originates
qv w ; q0v w 2 f0; 1g; v 2 N ; w 2 N ; v –w; ð21Þ
from all active super-links other than (i, j), must be at least
c. Note that for any super-node k, k – i, the outer-sum in xv ; yv 2 f0; 1g; v 2 N: ð22Þ
the denominator will contain at most one super-link from
k, because of (12). By (16), node v transmits to node w if and only if a super-
Constraints (13) are non-linear. A standard way of line- link containing v and w as transmitting and receiving
arizing these constraints (e.g., [35]) is to introduce a nodes, respectively, is active. Constraints (17) state that,
parameter, denote by M, and reformulate (13) by the fol- if v is transmitting, then another node w either is a recei-
lowing linear inequalities. The parameter M is chosen to ver, or experiences interference from v. The next set of con-
be large enough such that the inequality has no effect if straints ensures yw = 1 if qvw = 1 for any v. The SINR
zij = 0. conditions, one for each w 2 N , are formulated in (19).
These constraints can be linearized in a way similar to (15).
0 1
B C
X B X X C 5.3. Algorithm summary
B C
P v g v w P cB P u g uw zkl þ gC Mð1zij Þ;ði;jÞ 2 A;w 2 CðjÞ:
B C
v 2CðiÞ @ ðk;lÞ 2 A : u2CðkÞ A
The computational machinery that we propose for solv-
k–i;l–j
ing the routing and scheduling problem in a cooperative
ð15Þ network consists in the following steps.

Computationally, the number of SINR constraints (15) has Step 1. Define an initial subset S 0  S.
a great impact on the time required to solve the pricing Step 2. Solve the master problem of the LP relaxation of P
problem. Recall that the number of super-links jAj is expo- (Section 5.1).
nential in the size of the original graph G. In R1, there are Step 3. Solve the pricing problem using either of the two
P
ði;jÞ2A jCðjÞj SINR constraints. Below we present an alterna- models (Section 5.2).
tive model for the pricing problem using more variables Step 4. If a configuration s 2 S n S 0 with negative reduced
and inequalities but significantly less SINR constraints. cost is found, add s to S 0 and go to Step 2, other-
 wise go to the next step.
1 if node v 2 N is transmitting to w 2 N ;
qv w ¼ Step 5. Find an integer solution for the configurations in
0 otherwise: set S 0 .
 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1263

The last step can be performed by applying an integer

linear solver, or, if this takes excessive computing time, a
heuristic algorithm. In either case, P is not guaranteed to
be solved to global optimality, because some configura-
tions used at integer optimum may not be present in S 0
in this step. Ensuring integer optimality would require
embedding the above algorithm into a branch-and-price
scheme. Empirically, however, for all the network in-
stances used in our experiments, integer optimum is ob-
tained and verified (by the LP value, sometimes after
rounding), i.e., no branching is needed.
Among the steps in the algorithm, two are crucial to
computational efficiency. The first is the initial subset of
configuration S 0 , and the second is the solution of the
pricing problem. In order to generate S 0 , we use a Shortest
Path Allocation (SPA) heuristic that generates, for each
demand d 2 D, the shortest path from its source od to Fig. 6. Topology with density d = 0.14 and diameter D = 6, induced by
parameters P = 0.1 mW, g = 1010 mW, and c = 15.
its destination td, and adds to S 0 one configuration per
link that appears in at least one of the paths. The step
of solving the pricing subproblem is definitely the compu-
tational bottleneck. One approach to speed up this step is straint, are not connected in the original network. Fig. 5
not to solve the pricing problem to optimality (i.e., finding shows a small example of a network with six nodes and
the minimum reduced cost), but to terminate once a con- two demands h2, 4i and h4, 3i. Recall that, in this paper, a de-
figuration with negative reduced cost is found; this con- mand represents a single packet associated with a source
figuration is added to the master problem. Although the and a destination. The transmission power P = 0.2 mW, the
number of column generation iterations grows, the ap- thermal noise at the receiver g = 1010 mW, and the SNR
proach pays off as significantly less time is spent per threshold c = 10. Since node 4 is not connected to node 2
iteration. or node 3, the two demands cannot be satisfied without
cooperation. If we allow a 2-level of cooperation, the de-
mand h2, 4i is satisfied with the configurations 2 ? {3, 5}
6. Computational results
and {3, 5} ? 4, and the demand h4, 3i with the configura-
tions 4 ? {4, 6} and {4, 6} ? 3. Note that the second demand
We present first a set of illustrative examples to high-
uses the buffering link 4 ? {4, 6}.
light some insights into the performance gain of coopera-
The advantage of cooperation tends to decrease as the
tive relaying. Then, in the second subsection, we report
connectivity of the network increases. To quantify the con-
the bulk of computational results on a large set of random
nectivity, we use two network properties: (i) the network
instances.
density d ¼ N ðNL1Þ, that is the percentage of links that sat-
isfy the SNR constraint with respect to a fully connected
6.1. Illustrative examples network, and (ii) the network diameter D, that is, the max-
imum hop distance between any pair of nodes. Figs. 6 and
A clear advantage of cooperative relaying is to enable 7 show two networks that differ only in the level of
communication between nodes that, due to the SNR con- connectivity. In the first case the nodes transmit with
P = 0.1 mW. In the second case the power P = 0.4 mW. In
both cases, each pair of nodes i and j has a demand in both
directions, that is hi, ji and h j, ii. For the network in Fig. 6,
the optimal solution requires 130 time slots without coop-
eration, and only 110 time slots with cooperation, showing
a clear advantage of cooperation. In contrast, for the net-
work in Fig. 7 where the connectivity is much higher, the
optimal schedule length is 74, with or without
cooperation.
The traffic load, measured in the number of traffic de-
mands, is another important factor influencing the poten-
tial of cooperation. In particular, increasing the number
of traffic demands, the advantage of cooperative relaying
decreases. Fig. 8 shows an illustrative example of a grid
network with 12 nodes, P = 0.2 mW, g = 1010 mW, and
c = 10. We compare two problem instances: the first in-
stance has a single demand h1, 4i, the second instance
has two demands h1, 4i and h9, 12i. In the first case, coop-
Fig. 5. Cooperative networking to provide additional connectivity. erative relaying requires one time slot less to satisfy the
1264 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269

optimum is identical to that of conventional hop-by-hop

forwarding.

6.2. Computational evaluation

We have performed an assessment of the proposed

models by comparing the efficiency of cooperative relaying
to that of hop-by-hop forwarding on random network
topologies with up to 15 nodes. A first set of results consid-
ers networks with 13 nodes drawn on a square area of
100 m2. The thermal noise is set to g = 1010 mW and the
SNR threshold c = 10. The propagation gain is computed
as gij = (distanceij)3. The number of demands ranges from
3 (low traffic load) to 90 (high traffic load), while the trans-
mission power is either P = 0.1 mW or P = 0.4 mW. For
P = 0.1 mW, the SNR requirement results in sparse
Fig. 7. Topology with density d = 0.30 and diameter D = 2, induced by networks with D = 3 in average, while for P = 0.4 mW the
parameters P = 0.4 mW, g = 1010 mW, and c = 15. networks are denser with D 6 2 in average.
Figs. 9 and 10 report the cooperation gain averaged over
10 random instances. Cooperation gain is computed as the
demand. In the second case, cooperative networking has an
percentage difference between the optimal schedule
optimal solution where cooperation does not occur, i.e., the
lengths obtained with and without cooperation. Fig. 9

Fig. 8. The effect of traffic demand on cooperative networking.

 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1265

reports the cooperation gain in respect of the transmission mance gain is pretty small in value, and ranges between
power with a full demand matrix. Fig. 10 shows the perfor- 7.8% and 11.8%.
mance figure in the number of traffic demands with a fixed Table 1 gives additional details on the same set of
transmission power P = 0.1 mW. The results confirm the instances. For each instance we report the number of links
observations made for the illustrative examples in the jLj, the number of super-links jAj, and the computation
previous section. time in seconds. Note that, as expected, the computation
In particular, Fig. 9 shows that the cooperation gain time is much higher in the case of cooperation, and
decreases as network connectivity increases. Indeed, as increases with both the number of demands and the num-
explained in Section 6.1, when the transmission power ber of links (and thus super-links).
increases, the connectivity of the network also grows, since For 10 randomly generated instances, Figs. 11–13 show
more node pairs get their SNRs above the threshold. How- results for three power levels, P = 0.1 mW, P = 0.2 mW and
ever, the interference level increases as well, making 0.3 mW. The instances have 13 nodes and a full demand
cooperation less advantageous since it activates more matrix. The resulting three groups of network topologies
simultaneous transmitters. Moreover, the values of the are denoted by p1–p10, p11–p20, and p21–p30, respec-
cooperation gain that are basically upper limits of what tively. For each scenario, the corresponding figure reports
can be achieved in practice, are rather small and goes from the cooperation gain. These figures lead to another insight
6.1% to 9.9%. into the behavior of cooperative relaying, namely the very
Fig. 10 confirms that the cooperation gain decreases as large diversity in the performance gain. Indeed, very differ-
the number of traffic demands increases. The main reason ent results have been obtained with networks that share
for this behavior is again the interference level that goes up the same system parameters. For the three power levels,
with multiple parallel traffic relations. Moreover, the half- the ranges of the cooperation gain are [1.6%, 27.3%],
duplex operation prevents nodes from being involved [0%, 22.4%], and [0.6%, 19.4%], respectively.
simultaneously in different cooperation groups, and thus Tables 2–4 report detailed results for the instances in
reduces the advantage of cooperation with multiple Figs. 11–13. For each instance, the table shows the number
parallel transmissions of different packets. Also, the perfor- of links L, the number of super-links jAj, the number of
iterations of column generation, and the computation time
in seconds. We denote by LB1 and UB1 the lower and upper
bound obtained without cooperation, and LB2 and UB2 in
12% case of cooperation. The lower bounds are the LP optimum
values, and the upper bounds correspond to integer sche-
10%
|D| = "all"
8%
Coop. Gain

6% Table 1
A summary of the network scenarios and computational results.
4%
Power jDj No-cooperation Cooperation Gain (%)

2% jLj Time jAj Time

0.1 5 42.4 0.9 626.0 16.6 11.9
0% 0.1 10 47.2 2.2 683.2 123.0 8.2
0,1 0,2 0,3
0.1 20 41.6 2.5 609.6 147.4 7.8
Power 0.1 ‘‘all’’ 41.6 5.3 609.6 194.7 9.9
0.2 ‘‘all’’ 60.8 17.8 958.6 1792.7 7.8
Fig. 9. Cooperation gain versus transmission power with full traffic 0.3 ‘‘all’’ 76.4 23.1 1277.6 5781.8 6.1
demand matrix.

14%
30%
12%
Power = 0.1 25%
10% Power = 0.1
Coop. Gain

20%
Coop. Gain

8%
15%
6%
10%
4%

2% 5%

0% 0%
5 10 20 p1 p2 p3 p4 p5 p6 p7 p8 p9 p10
|D| Instances

Fig. 10. Cooperation gain versus number of traffic demands with fixed Fig. 11. Cooperation gain for 10 random instances with transmission
transmission power. power P = 0.1 mW.
1266 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269

30% reports results for networks with 15 nodes and the same
technical parameters as before. This time, the traffic de-
25% mands are generated in a specific way. The number of de-
Power = 0.2
mands jDj grows from 5 to 20. For each value of jDj, the
Coop. Gain

20%
source and destination nodes of the demands are chosen
15% with a fixed hop distance h, where h = 2, 3, 4, 5, 6. In prac-
tice, for h = 6, we have considered all the demands with
10% hop distance being equal to or greater than 6. The purpose
is to analyze the behavior of cooperation not only as a
5%
function of traffic load, but also as a function of source–
0% destination distance. The figure gives the percentage gain
p11 p12 p13 p14 p15 p16 p17 p18 p19 p20 of using cooperation in the schedule length. For example,
Instances for the instances with 20 demands and h P 6, the schedule
length by cooperation is 37% shorter than that without
Fig. 12. Cooperation gain for 10 random instances with transmission
power P = 0.2 mW. cooperation. The figure shows clearly that the higher is
the source–destination distance, the higher will be the
cooperation gain. This is due to the fact that in the cooper-
ation graph the hop distances between pairs of nodes are
30%
usually smaller, because of the presence of super-links that
25% give additional shortcuts.
Power = 0.3
20%
Coop. Gain

6.3. Additional comments on the results

15%
Recall that the proposed approach compares the opti-
10%
mal routing and scheduling for hop-by-hop forwarding
5% and cooperative relaying (decode-and-forward). Therefore,
the two mechanisms are compared considering their best
0% achievable performance when the propagation parameters
p21 p22 p23 p24 p25 p26 p27 p28 p29 p30
are all known. Since protocol overheads are not included,
Instances hop-by-hop forwarding becomes the special case of coop-
Fig. 13. Cooperation gain for 10 random instances with transmission
erative relaying where a single node is selected for each
power P = 0.3 mW. transmission. Thus the cooperation gain is always greater
than or equal to zero.
The values of the cooperation gain we have obtained for
dule lengths. The last column gives the cooperation gain, some illustrative examples and a set of random instances
computed as UB2UBUB
2
1
 100. point out that the advantage of cooperation is highly
Note that our column generation approach produces dependent on network topology and traffic pattern. In sin-
the global optimal integer solution whenever the ceiling gle source–destination pair scenarios, channel capacity
of the lower bound is equal to the upper bound. Inspecting improvements because of cooperation prevail. In network-
the tables, we observe that this is indeed the case for all the ing scenarios, the reuse level of radio resources is a key ele-
instances. ment, and the results show that the limitation of
Finally, Fig. 14 provides further insights into scenarios cooperation on the reuse level is particularly severe when
where the advantage of cooperation is apparent. The figure the network is dense and/or many traffic flows must be

Table 2
Performance evaluation of random topologies with 13 nodes in a square area of 1000 m2, with parameters P = 0.1 mW, c = 10, and g = 2  1010 mW. The time
is in seconds.

Ins. No cooperation Cooperation Gain (%)

jLj Iter Time LB1 UB1 jAj Iter Time LB2 UB2
p1 38 20 3.3 43.0 43 525 369 29.4 38.0 38 13.2
p2 38 27 6.3 324.3 325 526 609 309.5 303.8 304 6.9
p3 36 22 3.3 61.0 61 520 190 20.6 57.0 57 7.0
p4 42 26 5.7 115.7 116 593 329 151.0 112.0 112 3.6
p5 34 13 2.8 168.0 168 509 316 43.0 132.0 132 27.3
p6 34 21 3.5 125.0 125 510 127 42.9 122.5 123 1.6
p7 60 32 14.6 236.0 236 876 447 1093.4 216.5 217 8.8
p8 38 28 5.0 138.0 138 581 433 139.5 134.7 135 2.2
p9 50 8 3.3 95.0 95 741 43 43.9 93.0 93 2.2
p10 46 23 5.5 209.0 209 715 148 74.2 165.0 165 26.7
Ave. 41.6 5.3 609.6 194.7 9.93
 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1267

Table 3
Performance evaluation of random topologies with 13 nodes in a square area of 1000 m2, with parameters P = 0.2 mW, c = 10, and g = 2  1010 mW. The time
is in seconds.

Ins. No cooperation Cooperation Gain (%)

jLj Iter Time LB1 UB1 jAj Iter Time LB2 UB2
p11 50 23 5.1 73.0 73 786 171 77 73.0 73 0.00
p12 54 19 7.1 290.0 290 812 798 1288 236.3 237 22.36
p13 50 34 12.4 232.0 232 696 1189 1085 215.3 216 7.41
p14 60 23 11.3 148.0 148 1004 399 639 140.3 141 4.96
p15 54 25 9.8 257.0 257 845 546 834 230.3 231 11.26
p16 56 32 12.8 85.8 86 885 292 377 85.7 86 0.00
p17 84 78 75.6 162.6 163 1396 736 7200 161.0 161 1.24
p18 64 26 14.1 203.0 203 1020 1092 3561 189.5 190 6.84
p19 74 19 11.8 225.0 225 1168 251 1304 207.0 207 8.70
p20 62 37 17.7 236.3 237 974 610 1562 205.0 205 15.61
Ave. 60.8 17.8 958.6 1793 7.84

Table 4
Performance evaluation of random topologies with 13 nodes in a square area of 1000 m2, with parameters P = 0.3 mW, c = 10, and g = 2  1010 mW. The time
is in seconds.

Ins. No-cooperation Cooperation Gain (%)

jLj Iter Time LB1 UB1 jAj Iter Time LB2 UB2
p21 70 38 26.79 150.5 151 1057 1365 7200 149 149 1.34
p22 64 29 16.71 184 184 1014 1848 7061 166.4 167 10.18
p23 52 31 11.48 230 230 820 1489 2026 203.5 204 12.75
p24 80 40 41.37 186.22 187 1424 861 6479 183.2 184 1.63
p25 80 22 16.16 200.5 201 1309 778 7200 193.5 194 3.61
p26 68 36 29.21 181.75 182 1068 461 2169 177 177 2.82
p27 98 42 30.88 161.33 162 1850 421 7200 154 154 5.19
p28 82 28 20.41 162.68 163 1337 736 7200 162 162 0.62
p29 92 34 25.27 198 198 1630 273 4083 191 191 3.66
p30 78 18 12.54 209 209 1267 1176 7200 174.3 175 19.43
Ave. 76.4 23.1 1277.6 5782 6.12

served in parallel. This is due to the high level of interfer- to a higher degree of freedom in optimizing scheduling
ence that makes cooperation less attractive and the fact and routing, and the advantage of increased transmission
that nodes cannot be in multiple cooperation groups at range by cooperation becomes more apparent.
the same time because of half duplex operation.
The results yields another insight that could not be ob-
served with classical schemes that consider two-steps 7. Conclusions
(hops) cooperation only. Namely, the cooperation gain in-
creases remarkably with the number of hops traversed In this paper we have proposed an optimization ap-
by traffic flows. This is because a large hop distance leads proach for evaluating the resource efficiency of cooperative
relaying vis-a-vis conventional multi-hop forwarding. We
have modeled interference through the physical model
based on SINR values, and considered the decode-and-for-
ward cooperation scheme that defines a cooperation graph.
Considering a spatial time division multiplexing frame-
work, we have designed a methodology for optimizing
scheduling and routing both in case of cooperative relay
and traditional multi-hop forwarding. The resource effi-
ciency has been measured considering the minimum num-
ber of time slots required to serve all traffic demands.
The proposed methodology extends previous works in
allowing for evaluating the efficiency of cooperation in net-
works with arbitrary topology, where the interference is
not just statistically characterized but controlled through
the scheduling of simultaneous transmissions. Moreover,
our approach provides a tool for managing in an efficient
Fig. 14. Cooperation gain versus the number of hops for different values way the radio resource in networks of cooperative nodes
of jDj. through joint routing and scheduling optimization.
1268 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269

The results presented show that the overall efficiency Conference on Wireless Networks, Communications and Mobile
Computing, 2005.
greatly depends on the network topology and the traffic
[19] J. Boyer, D.D. Falconer, H. Yanikomeroglu, Multihop diversity in
pattern. The limitation of cooperation on improving the re- wireless relaying channels, IEEE Transactions on Communications
use level of radio resources is particularly apparent when 52 (10) (2004) 1820–1830.
the network is dense and a large number of traffic flows [20] K. Azarian, H.E. Gamal, P. Schniter, On the achievable diversity-
multiplex tradeoff in half-duplex cooperative channels, IEEE
must be served. For scenarios where the underlying net- Transactions on Information Theory 51 (12) (2005) 4152–4172.
work has many direct links as well as source–destination [21] T. Abe, H. Shi, T. Asai, H. Yoshino, A relaying scheme for MIMO
pairs, the gain of cooperation tends to diminish. On the wireless networks with multiple source and destination pairs, in:
Proceedings of 11th Asia-Pacific Conference on Communications
other hand, the advantage of cooperation in a networking (APCC), 2005.
context is significant as the number of hops increases. [22] G. Li, H. Liu, On the capacity of broadband relay networks, in:
Proceedings of 38th Asilomar Conference on Signals, Systems and
Computers, 2004.
Acknowledgments [23] G. Scutari, S. Barbarossa, D. Ludovici, Cooperation diversity in
multihop wireless networks using opportunistic driven multiple
access, in: Proceedings of IEEE Workshop on Signal Processing
We are grateful to Lazzaro Baj for his contribution to Advances in Wireless Communication (SPAWC03), June 2003.
the experimental evaluation. The work of the first two [24] A. Bletsas, H. Shin, M.Z. Win, Cooperative communications with
outage-optimal opportunistic relaying, IEEE Transactions on
authors has been partially supported by project PRIN SES- Wireless Communication 6 (9) (2007) 3450–3460.
AME of the Italian ministry of university and research, and [25] M. Qin, R. Blum, Capacity of wireless ad hoc networks with
the European Network of Excellence EURO-NF. The work of cooperative diversity: a warning on the interaction of relaying and
multi-hop routing, in: Proceedings of IEEE ICC, 2005.
the last author has been partially supported by the Swed- [26] P. Gupta, P.R. Kumar, The capacity of wireless networks, IEEE
ish ELLIIT network. We thank the anonymous reviewers Transactions on Information Theory 46 (2) (2000) 388–404.
for their valuable comments and suggestions [27] A.H. Madsen, Capacity bounds for cooperative diversity, IEEE
Transactions on Information Theory 52 (4) (2006) 1522–1544.
[28] I. Krikidis, J.S. Thompson, S. McLaughlin, N. Goertz, Max–min relay
selection for legacy amplify-and-forward systems with interference,
References IEEE Transactions on Wireless Communications 8 (6) (2009) 3016–
3027.
[1] M. Abolhasan, T. Wysocki, E. Dutkiewicz, A review of routing [29] Y. Zhu, H. Zheng, Understanding the impact of interference on
protocols for mobile ad hoc networks, Ad Hoc Networks 2 (1) (2004). collaborative relays, IEEE Transactions on Mobile Computing 7 (6)
[2] M. Gastpar, M. Vetterli, On the capacity of wireless networks: the (2008) 724–736.
relay case, in: Proceedings of IEEE INFOCOM, New York, US, 2002. [30] P. Björklund, P. Värbrand, Di Yuan, A column generation method for
[3] A. Sendonaris, E. Erkip, B. Aazhang, User cooperation diversity, part I: spatial TDMA scheduling in ad hoc networks, Ad Hoc Networks 2 (4)
system description/part II: implementation aspects and performance (2004) 405–418.
analysis, IEEE Transactions on Communications 51 (11) (2003) [31] D.M. Blough, S. Das, G. Resta, P. Santi, A framework for joint
1927–1948. scheduling and diversity exploitation under physical interference in
[4] E.C. van der Meulen, Three-terminal communication channels, wireless mesh networks, in: Proceedings of IEEE MASS, 2008.
Advanced Applied Probability 3 (1971) 120–154. [32] A. Capone, G. Carello, I. Filippini, S. Gualandi, F. Malucelli, Solving a
[5] T.M. Cover, A.A.E. Gamal, Capacity theorems for the relay channel, resource allocation problem in wireless mesh networks: a
IEEE Transactions on Information Theory 25 (5) (1979) 572–584. comparison between a CP-based and a classical column
[6] M. Dohler, F. Said, H. Aghvami, Concept of virtual antenna arrays, in: generation, Networks 55 (3) (2010) 221–233.
Proceedings of IEEE Globecom 2002, Taipei, Taiwan, 2002. [33] A. Capone, G. Carello, I. Filippini, S. Gualandi, F. Malucelli, Routing,
[7] M. Dohler, A. Gkelias, H. Aghvami, 2-Hop distributed MIMO scheduling and channel assignment in wireless mesh networks:
communication system, IEEE Electronics Letters, 2003. optimization models and algorithms, Ad Hoc Networks 8 (6) (2010)
[8] J.N. Laneman, D.N.C. Tse, G.W. Wornell, Cooperative diversity in 545–563.
wireless networks: efficient protocols and outage behavior, IEEE [34] L.A. Wolsey, Integer Programming, John Wiley & Sons, New York,
Transactions on Information Theory 50 (12) (2004) 3062–3080. 1998.
[9] P. Herhold, E. Zimmermann, G. Fettweis, Cooperative multi-hop [35] H.P. Williams, Model Building in Mathematical Programming, third
transmission in wireless networks, Computer Networks 49 (2005) ed., Wiley, 1995.
299–324.
[10] M. Gastpar, G. Kramer, P. Gupta, The multiple-relay channel: coding
and antenna-clustering capacity, in: Proceedings of IEEE
Internaional Symposium on Information Theory, Lausanne, Antonio Capone is an Associate Professor at
Switzerland, 2002. the Information and Communication Tech-
[11] A. Hst-Madsen, J. Zhang, Capacity bounds and power allocation for nology Department (Dipartimento di Elett-
wireless relay channels, IEEE Transactions on Information Theory 51 ronica e Informazione) of the Technical
(6) (2005) 2020–2040. University of Milan (Politecnico di Milano),
[12] G. Kramer, M. Gastpar, P. Gupta, Cooperative strategies and capacity where he is the director of the Advanced
theorems for relay networks, IEEE Transactions on Information Network Technologies Laboratory (ANTLab).
Theory 51 (9) (2005) 3037–3063. Dr. Capone is co-founder and CTO of Mobi-
[13] A. Stefanov, E. Erkip, Cooperative coding for wireless networks, in: MESH (www.mobimesh.eu), a spin-off com-
Proceedings of IEEE Conference on Mobile Wireless Communication pany of Politecnico di Milano. His expertise is
Networks, Stockholm, Sweden, 2002.
on networking and his main research activi-
[14] M.C. Valenti, B. Zhao, Distributed turbo coded diversity for the relay
ties include protocol design (MAC and rout-
channel, IEE Electronics Letters 39 (10) (2003) 786–787.
[15] M.C. Valenti, B. Zhao, Distributed turbo codes: towards the capacity ing) and performance evaluation of wireless access and multi-hop
of the relay channel, in: Proceedings of IEEE VTC 2003. networks, traffic management and quality of service issues in IP net-
[16] B. Zhao, M.C. Valenti, Practical relay networks: a generalization of works, and network planning and optimization. On these topics he has
hybrid-ARQ, IEEE Journal on Selected Areas in Communications 23 published more than 130 peer-reviewed papers in international journal
(1) (2005) 7–18. and conference proceedings, and holds five patents. He received the M.S.
[17] X. Deng, A.M. Haimovich, Power allocation for cooperative relaying and Ph.D. degrees in electrical engineering from the Politecnico di Milano
in wireless networks, IEEE Communications Letters 9 (11) (2005) in 1994 and 1998, respectively. In 2000 he was visiting professor at UCLA,
994–997. Computer Science department. He currently serves as editor of ACM/IEEE
[18] S. Serbetli, A. Yener, Optimal power allocation for relay assisted Trans. on Networking, Wireless Communications and Mobile Computing
F/TDMA ad hoc networks, in: Proceedings of International
 A. Capone et al. / Ad Hoc Networks 9 (2011) 1256–1269 1269

(Wiley), Computer Networks (Elsevier), and Computer Communications Di Yuan received his M.Sc. degree in com-
(Elsevier). He was guest editor of a few journal special issues and served puter science and engineering, and Ph.D.
in the technical program committee of major international conferences degree in operations research at Linköping
(including Mobicom, INFOCOM, SECON, MASS, Globecom, ICC, LCN, Net- Institute of Technology in 1996 and 2001,
working, WoWMoM), as Technical Program Chair of Ifip MEDHOCNET respectively. At present he is full professor in
2006, Poster & Demo co-chair of SECON 2009, Workshop chair of INFO- telecommunications at the Department of
COM 2010, and publicity chair of MOBIQUITOUS 2007. He is currently Science and Technology, Linköping University.
involved in the scientific and technical activities of several national and His research interests span design, analysis,
European research projects, and he leads several industrial projects. He is and resource optimization of telecommuni-
a Senior Member of the IEEE. cation systems. He has published over 70
refereed articles in international journals and
conferences. His current research mainly
Stefano Gualandi obtained the Dr. Ing. addresses network design and bandwidth allocation in 3G/4G systems,
Degree in Computer Science Engineering and resource management in ad hoc networks.
summa cum laude in 2002. He got a Master in
Artificial Intelligence in 2004, at the Kath-
olieke Universiteit te Leuven, Belgium. He got
a Ph.D. from Politecnico di Milano, Italy, in
2008. At present he is a postdoctoral fellow at
the Department of Mathematics, University of
Pavia, Italy, working on algorithm engineering
for network problems. His research interests
focus on the models and algorithms for com-
binatorial optimization problems, integrating
techniques from Operations Research and Constraint Programming.