EE-OLSR: ENERGY EFFICIENT OLSR ROUTING PROTOCOL FOR MOBILE AD-HOC

NETWORKS

Floriano De Rango
Marco Fotino
Salvatore Marano
D.E.I.S. Department, University of Calabria, Via P.Bucci, Rende, Italy
{derango, mfotino, marano}@deis.unical.it

ABSTRACT perspectives ([3]-[7]). These protocols try to satisfy

various properties, like: distributed implementation,
This paper presents two novel mechanisms for the OLSR efficient utilization of bandwidth and battery capacity,
routing protocol, aiming to improve its energy optimization of metrics (like throughput and end-to-end
performance in Mobile ah-hoc Networks. Routing delay), fast route convergence and freedom from loops.
protocols over MANET are an important issue and many In this work, we will try to analyze the performance of a
proposals have been addressed to efficiently manage MANET from the energy point of view. Since mobile
topology information, to offer network scalability and to hosts today are powered by battery, efficient utilization of
prolong network lifetime. However, few papers consider battery energy is very important. Battery life, therefore,
a proactive protocol (like OLSR) to better manage the can also affect the overall network communication
energy consumption. OLSR presents the advantage of performance: when a node exhausts its available energy,
finding a route between two nodes in the network in a it ceases to function and the lack of mobile hosts can
very short time, thanks to its proactive scheme, but it can result in network partitioning. For that reason, reducing
expend a lot of resources selecting the MultiPoint Relays power consumption is an important issue in ad hoc
(MPRs) and exchanging Topology Control information. wireless networks. However, the majority of the routing
We propose a modification in the MPR selection proposals to date have not focused on the power
mechanism of OLSR protocol, based on the Willingness constraints of unethered nodes: traditional routing
concept, in order to prolong the network lifetime without protocols tend to use shortest path algorithms (minimum
losses of performance (in terms of throughput, end-to-end hop count) without any consideration of energy
delay or overhead). Additionally, we prove that the consumption, often resulting in rapid energy exhaustion
exclusion of the energy consumption due to the for the small subset of nodes in the network that
overhearing can extend the lifetime of the nodes without experience heavy traffic loads. In recent years a number
compromising the OLSR functioning at all. A comparison of power-aware metrics have been proposed (like [11]-
of an Energy-Efficient OLSR (EE-OLSR) and the [14]). The majority of these metrics has been applied to
classical OLSR protocol is performed, testing some DSR routing protocol, so we decided to perform an
different well-known energy aware metrics such as energetic evaluation of another protocol, i.e. the proactive
MTPR, CMMBCR and MDR. We notice how EE-OLSR protocol OLSR, arrived to the RFC status. In particular
outperforms classical OLSR, and MDR confirms to be the the energy behavior of OLSR protocol has been
most performing metric to save battery energy in a dense evaluated and a novel energy aware MultiPoint Relay
mobile network with high traffic loads. selection mechanism has been proposed. We want to
investigate the effects of applying energy-aware routing
to the OLSR protocol in a MANET, to evaluate the
INTRODUCTION influence of overhearing and idle activity on the energy
In the last few years, thanks to the proliferation of consumption in a network using the IEEE 802.11
wireless devices, the use of mobile networks is growing technology and to check if these considerations could
very fast. In particular, a very large number of recent affect the performance of a protocol that ensures a good
studies focused on Mobile Ad-hoc Networks (MANETs) QoS in terms of end-to-end delay.
[1]. The performance of a mobile ad hoc network This paper is organized as follows: section II presents a
depends on the routing scheme employed, and the short summary on some related works about energy-
traditional routing protocols do not work efficiently in a aware metrics for routing protocols; OLSR is briefly
MANET. This kind of network, in fact, has a dynamic introduced in section III; our mechanisms to improve
topology (every node can move randomly and the radio energy consumption in OLSR (with energy-aware
propagation conditions change rapidly over the time) and selection and overhearing exclusion) is explained in
a limited bandwidth (so that the control traffic overhead section IV; finally, performance evaluation of different
must be reduced to the minimum) [2]. Developing energy metrics and conclusions are summarized in the
routing protocols for MANETs has been an extensive last two sections.
research area in recent years, and many proactive and
reactive protocols have been proposed from a variety of

1
978-1-4244-2677-5/08/$25.00©2008 IEEE
 RELATED WORKS as long as all nodes in a route have sufficient remaining
The routing protocols for mobile networks can be energy (over a threshold) and the latter when all routes to
classified into four broad categories: proactive, reactive, destination have at least a node with less energy than the
hybrid, and cluster-based. These protocols try to satisfy threshold.
various properties to reach the best compromise in term
of scalability, mobility support, and energy consumption. C. MDR (Minimum Drain Rate)
The need of energy efficiency is a problem concerning
with the constraints imposed by battery capacity and heat Power saving mechanisms based only on the remaining
dissipation which are opposed by the desire of power cannot be used to establish the best route between
miniaturization and portability. In a wireless network, we source and destination nodes. If a node is willing to
have different opportunities to increase energy efficiency. accept all route requests only because it currently has
One of them is the possibility of dynamically offloading enough residual battery capacity, too much traffic load
computation from the local terminal to remote, energy- will be injected through that node. In this sense, the
rich nodes (e.g., fixed servers). Another chance comes actual drain rate of power consumption of the node will
from making various network protocols, such as link, tend to be high, resulting in an unfair sharp reduction of
MAC routing and transport protocols, energy-aware. In battery power. To address the above problem, the
recent years a number of power-aware metrics have been Minimum Drain Rate (MDR) [11] mechanism can be
proposed at the network layer (like [2]-[3], [6]-[7],[11]- utilized with a cost function that takes into account the
[14],[16],[17]). Here we present a brief description of the drain rate index (DR) and the residual battery power
most relevant power-aware routing metrics proposed (RBP) to measure the energy dissipation rate in a given
recently. node. In this mechanism, the ratio  ⁄ , at node  ,
indicates when the remaining battery of node  will be
A. MTPR (Minimum Total Transmission Power exhausted, i.e., how long node  can keep up with
Routing) and MBCR (Minimum Battery Cost routing operations with current traffic conditions. The
Routing) corresponding cost function can be defined as:  
 ⁄ . Therefore, the maximum lifetime of a given
The MTPR [2] mechanism uses a simple energy metric, path ! is determined by the minimum value of  over
represented by the total energy consumed to forward the the path. Finally, the MDR mechanism is based on
information along the route. This way, MTPR reduces the selecting the route " , contained in the set of all possible
overall transmission power consumed per packet, but it routes  between the source and the destination , having
does not affect directly the lifetime of each node (because the highest maximum lifetime value.
it does not take account of the available energy of
network nodes). Notice that, in a fixed transmission
power context, this metric corresponds to a Shortest Path OPTIMIZED LINK-STATE ROUTING (OLSR)
routing. The Optimized Link State Routing (OLSR) protocol is an
Let
  be the battery capacity of node  at time t. We optimization of the classical link state algorithm, adapted
define   as a battery cost function of node  . The less to the requirements of a MANET ([10]). Because of their
capacity a node has, the more reluctant it is to forward quick convergence, link state algorithms are somewhat
packets; the proposed value is    1⁄
 . The less prone to routing loops than distance vector
metric that minimizes this function to forward a packet is algorithms, but they require more CPU power and
called MBCR. memory. They can be more expensive to implement and
support and are generally more scalable. OLSR operates
B. MMBCR (Min-Max Battery Cost Routing) and in a hierarchical way (minimizing the organization and
CMMBCR (Conditional MMBCR) supporting high traffic rates). The key concept used in
OLSR is that of multipoint relays (MPRs). MPRs are
If only the summation of battery costs on a route is selected nodes which forward broadcast messages during
considered, a route containing nodes with little remaining the flooding process. This technique substantially reduces
battery capacity may still be selected. MMBCR ([2]), the message overhead as compared to a classical flooding
defines the route cost as:     max   . The mechanism (where every node retransmits each message
desired route  is obtained so that received). This way a mobile host can reduce battery
    min   , where  is the set of all consumption. In OLSR, link state information is
possible routes. Because MMBCR considers the weakest generated only by nodes elected as MPRs. An MPR node
may choose to report only links between itself and its
and crucial node over the path, a route with the best
condition among paths impacted by each crucial node MPR selectors. Hence, contrarily to the classical link
over each path is selected. state algorithm, partial link state information is
CMMBCR metric ([2]) attempts to perform a hybrid distributed in the network. This information is then used
approach between MTPR and MMBCR, using the former for route calculation. OLSR provides optimal routes (in

2
terms of number of hops). The protocol is particularly
suitable for large and dense networks as the technique of
MPRs works well in this context.

Figure 2. Simulation trace scansion and states sequences

individuation.

Fig. 2. The EA-Willingness Setting heuristic

Table 1. Energy-based Willingness Selection.

Figure 1. MPR election in OLSR protocol.
Battery
Low Medium High
ENERGY EFFICIENT OLSR (EE-OLSR) Lifetime
Short W_LOW W_LOW W_LOW
With EE-OLSR (Energy-Efficient OLSR) we denote a
Medium W_LOW W_DEFAULT W_DEFAULT
routing protocol obtained modifying OLSR in order to
Long W_DEFAULT W_HIGH W_HIGH
improve its energy behavior, without loss of
performance. In addition to the energy-aware metrics
presented in the last section, we studied two ad-hoc The implementation of EA-Willingness Setting heuristic
mechanisms that will be shown in this section: the EA- was obtained implementing the following pseudo-code in
Willingness Setting and the Overhearing Exclusion. the OLSR protocol:

A. EA-Willingness Setting mechanism

The Energy Aware Willingness Setting is a mechanism to

involve energetic considerations in MPR selection. The
OLSR specification has a variable, the “willingness” of a
node, representing the availability of that node to act as a
MPR for its neighbors. By default, each node declares a
default willingness value.
In EE-OLSR, each node, calculating its own energetic
status, can declare an appropriate willingness. We
decided to base the willingness selection on both metrics:
the battery capacity and the predicted lifetime (based on
the energy-drain rate) of a node. The heuristic used to
associate a willingness (“default”, “low” or “high”) to a
pair (battery, lifetime) is shown in Fig. 2 and in Tab. 1.
For example, in condition of high battery value, if the In our implementation, we decided to use the ratio
predicted lifetime is short a node declares a between actual and initial energy of a node to measure its
W_DEFAULT willingness. On the other hand, if a longer battery capacity (with values between 0.0 and 1.0) and to
node lifetime is predicted (because the node is measure the predicted lifetime of a node in seconds
experimenting low traffic), the node can declare a (considering 65535 as infinity, for representation
W_HIGH willingness. In the same way, if the battery reasons). We chose to consider less than 10% of residual
charge is low a node is less available to become MPR and capacity as low battery values, and we stated that less of
declares a W_LOW willingness value (whatever lifetime 10 seconds predicted are a short lifetime, while more than
it predicts). This permits a better load balancing to be 100 second are a long lifetime.
obtained and node with lower residual energy are not
stressed. All willingness states are listed in table below.

3
B. Overhearing Exclusion packet at node 5 and to overhear the packet,
respectively. N represents the average number of
Another mechanism that allows energy saving in OLSR neighboring nodes affected by a transmission from node
protocol (without changing its behavior) is the 4 . Equation 1 implies that when the network is denser,
Overhearing Exclusion. Turning off the device when a packet overhearing causes more energy consumption.
unicast message exchange happens in our neighborhood,
can save a large amount of energy. This can be achieved B. Simulation Parameters
using the signalling mechanisms of the lower layers (i.e.
the RTS/CTS exchange performed by IEEE 802.11 to To evaluate EE-OLSR protocol, we used the ns-2
avoid collisions), and do not affect the protocol network simulator. We simulated a dense wireless
performance. In facts, OLSR does not takes any network, with 50 nodes moving in a 870 × 870 m area
advantage from unicast network information directed to (with a density of about 66 nodes/km2). Each node moves
other nodes (while other protocols such as DSR have a randomly in this area, with a speed of 3 m/s and no pause
mechanism to do so). We will be able to show, via time. Between mobile hosts there are 12 CBR/UDP
simulation, the amount of energy saved with this sources generating 20 packets/s (with a packet size of 512
mechanism. bytes). The duration of each simulation is 400 seconds
(with a setup time at the beginning, without traffic). To
C. Energy-aware Packet Forwarding extract average values, we simulated each scenario 5
times. NS-2 simulator allows to extract from a simulation
After the MPR election it is important to select the next many interesting parameters, like throughput, data packet
hop for data packet forwarding (among the MPR delivery ratio, end-to-end delay and overhead. To have
neighbors set). For this purpose we decided to consider detailed energy-related information over a simulation, we
some energy aware metrics such as explained in section modified the ns-2 code to obtain the amount of energy
II. These different metrics present some advantages and consumed over time by type (energy spent in
drawbacks, as explained in [2] . This approach permitted transmitting, receiving, overhearing or in idle state). This
us to decouple the MPR election process from route way, we obtained accurate information about energy at
selection mechanism. MPTR, MMBCR and MDR have every simulation time. We used these data to evaluate the
been considered, as shown in the following section. protocols from the energetic point of view: we will see
the impact of each protocol on different new parameters,
PERFORMANCE EVALUATION like the number of nodes alive over time (to check the
In this section we illustrate the energy consumption lifetime of nodes), the expiration time of connections (to
model adopted, the simulation parameters applied and see the network lifetime), and the energy usage divided
simulation results. by type (receiving, transmitting, overhearing). All used
simulation parameters are listed in Tab. 2 below:
A. Energy Consumption Model

We assume all mobile nodes to be equipped with IEEE Table 2. Simulation parameters.
802.11g network interface card, with data rates of 54 Modulation QPSK
Mbps. The energy needed to transmit a packet p from Area 870m x 870m
node  is: #$% &,    ( · * · ! Joules, where i is the
Nodes 50
current (in Ampere), v the voltage (in Volt), and ! the
Nodes speed 3 m/s
time taken to transmit the packet p (in seconds). In our
Simulation Time 400 s
simulations, the voltage v is chosen as 5 V and we
Traffic Sources 12
assume that the packet transmission time ! is calculated
Traffic Type CBR
by &, ⁄6 · 10/  0 &1 ⁄54 · 10/  seconds, where &, is
Packet Size 512 bytes
the packet header size in bits and &1 the payload size. As
shown in [8], we assume the energy consumption caused Start of Traffic 30 s
by overhearing a packet is the same as the energy End of Traffic 380
consumed by actually receiving the packet. The energy Transmission Power 1.4 W
#&, 4  consumed to transmit a packet from node 4 to Reception Power 1.0 W
node 5 is given by: Idle Power 0.0 W

#&, 4   #$% &, 4  0 #% &, 5  0 6 7 1 (1) C. Simulation Results

· # &,  
Many simulations have been assessed in order to test the
where #$% , #% , and # denote the amount of energy energy consumption by type, the energy-aware
spent to transmit the packet from node 4 , to receive the

4
willingness mechanism and the impact of different We can see how the use of an energy-aware willingness
energy-aware metrics on the protocol performance. Let selection can extend the lifetime of network nodes (and,
EE-OLSR with energy-aware metrics for packet thus, of connections) of several seconds. Most
forwarding be EE-OLSR-MPTR, EE-OLSR-MMBCR connections last about 10 seconds more, and the last
and EE-OLSR-MDR for, respectively, the MPTR, connection to expire dies about 60 second later.
MMBCR and MDR metrics. The standard version of The prolonged lifetime of nodes and connections
OLSR with minimum hop count metric is indicated positively affects the throughput: while the classical
simply with OLSR term. OLSR begins to lose data because of the lack of nodes,
the use of EA-Willingness Setting can improve the
1) Influence of Idle Power consumption network performance.

Fig. 3 shows the amount, in a log-scaled percentage, of

energy consumption in a simulation, by type. This figure 120000

Throughput [bytes/sec]
illustrates that the larger part of energy is spent in idle 100000 w EA-Will.
state (when the node is not using its network device): this w/o EA-Will.
state absorbs about 90% of the energy consumption of 80000
mobile devices. To clean results from this value 60000
(dominant as well as almost protocol independent), we
will ignore the idle power consumption in following 40000
simulations. 20000
0
30 80 130 180
Time [s]

Figure 5. Network Throughput, with and without EA-

Willingness Setting.

3) Overhearing Exclusion

In Fig. 3 we could note how overhearing consumes

nodes’ energy more than useful types of consumption
(transmission and reception power). Fig. 6 shows clearly
how much the use of a mechanism of Overhearing
Exclusion can improve the energetical performance of a
Figure 3. Energy consumption by type, in percentage (log mobile network.
scale). 6
Average Node Energy [J]

2) EA-Willingness Setting 5
4
To show how the EA-Willingness Setting mechanism
improves the performance of a MANET using OLSR, we 3
plotted the expiration time of connections and the 2 w/o Overhear.
network aggregate throughput, in figures 4 and 5. w Overhear.
1
180 0
w/o EA-Will.
Expiration Time [s]

160 0 100 200 300 400

w EA-Will. Time [s]
140
120
Figure 6. Average Node Energy by Time, with and
100 without Overhearing Exclusion.
80
60 In facts, without overhearing energy consumption, the
energy in the network is consumed very slowly, allowing
1 2 3 4 5 6 7 8 9 10 11 12
the nodes to send and receive packets for a longer time.
Connections

Figure 4. Expiration Time of Connections, with and

without EA-Willingness Setting.
4) EE-OLSR under different metrics versus OLSR

5
The following simulations will compare classical OLSR OLSR
protocol with EE-OLSR (using both EA-Willingness and 120000

Throughput [bytes/sec]
EEOLSR-MTPR
Overhearing Exclusion mechanisms), with some of the 100000
different energy-aware metrics proposed in literature. EEOLSR-MDR
80000
EEOLSR-CMMBCR
300 60000
OLSR
250 EEOLSR-MTPR 40000
Expiration Time [s]

EEOLSR-MDR 20000
200
150 0
30 80 130 180 230 280
100 Time [s]
50
Figure 9. Network Throughput, with classic OLSR and
0 different energy-aware metrics applied to EEOLSR.
1 2 3 4 5 6 7 8 9 10 11 12
Connections In Fig. 9 we plotted the network throughput over the
time: while OLSR throughput falls down after about 50
Figure 7. Expiration Time of Connections, with classic seconds; EE-OLSR delivers packets until the end of
OLSR and different energy-aware metrics applied to simulations (whatever metric. is used) We can note the
EEOLSR. particularly good performance of MDR metric.
Finally, Tab. 3 shows an overview of the performance of
The Fig. 7 shows the expiration time of connections. different simulated protocols: EE-OLSR outperforms
With every metric applied, EE-OLSR outperforms classical OLSR (especially in association with MDR
classical OLSR, especially using the MDR metric. metric), delivering more packets to destination, and
As shown in Fig. 8, EE-OLSR also guarantees a longer extending the lifetime of nodes and connections.
lifetime for every node in the network. While with Moreover, the use of mechanisms and metrics of EE-
classical OLSR all nodes tend to exhaust their energy OLSR does not lead to any loss of performance in terms
almost at the same time, EE-OLSR extends their lifetime of end-to-end delay, protocol overhead or path length. It
up to the end of the simulations: in particular, using is possible to observe the increase in the delivered
MDR metrics, about 20 nodes can survive over 400 packets and this is due to the longer nodes lifetime that
seconds. permits to support more traffic. Moreover, according with
Fig.9, the throughput increases producing a lower end-to-
50 end data packet delay.
Number of Alive Nodes

40
Table 3. Performance comparison of classic OLSR and
30 different energy-aware metrics applied to EE-OLSR.
EEOLSR-MTPR OLSR EEOLSR EEOLSR EEOLSR
20
EEOLSR-MDR MTPR CMMBCR MDR
EEOLSR-CMMBCR Packets 11811 26458 28453 32672
10
OLSR Delivered
E2E Delay 4.1940 4.0971 4.1800 3.8974
0 (msec)
30 130 230 330 Normalized 0.1526 0.1655 0.1558 0.1551
Overhead
Time [s] Average Hop 2.27 2.21 2.28 2.25
Count
Figure 8. Expiration Time of Nodes, with classic OLSR Node Average 149.79 274.46 276.23 302.56
and different energy-aware metrics applied to EEOLSR. Lifetime (s)
Mean 81.69 147.41 154.03 171.24
Connection
Expiration
Lifetime (s)

6
 [16] M.Fotino, A.Gozzi, F.De Rango, S.Marano, J.-C.Cano, C.Calafate,
P.Manzoni, “Evaluating Energy-aware Behavior of Proactive and
Reactive Routing Protocols for Mobile Ad Hoc Networks,” in 10th
CONCLUSIONS Int.Symposium on Performance Eval.of Computer and
Telecommunication Systems (SPECTS’07), 16-18 July, San Diego, CA,
A novel energy aware MPR election policy has been USA.
proposed. This novel features allows energy node to be [17] F. De Rango, et al., “OLSR vs DSR: A comparative analysis of proactive
preserved for longer time. A traffic load balancing and reactive mechanisms from an energetic point of view in wireless ad
hoc networks,” pubblished on-line 28 may 2008 on Computer
between MPR nodes has been achieved and performance Communication Journal.
improvement of OLSR in comparison with OLSR based
on minimum-hop count has been obtained. More energy-
aware metrics have been evaluated and the MDR resulted
the best choice in the MPR election and route selection
between source and destination. EE-OLSR outperforms
OLSR in terms of throughput, average nodes lifetime,
connection expiration time, nodes lifetime, preserving the
normalized control overhead. In future works EE-OLSR
will be compared with MDR based DSR in order to
evaluate the energy behaviour of two different topology
management strategies.

REFERENCES
[1] Internet Engineering Task Force, "Manet working group charter",
http://www.ietf.org/html.charters/manet-charter.html.
[2] C.-K. Toh "Maximum Battery Life Routing to Support Ubiquitous
Mobile Computing in Wireless Ad Hoc Networks", in IEEE
Communication Magazine, June 2001.
[3] E.Royer and C.-K. Toh, "A Review of Current Routing Protocols for Ad
Hoc Mobile Wireless Networks", IEEE Personal Communications
Magazine, Vol.6, No. 2, April 1999.
[4] C.E. Perkins, E.M. Royer, "Ad-hoc On-Demand Distance Vector
Routing", Proceedings of the 2nd IEEE Workshop on Mobile Computing
Systems and Applications, Feb. 1999.
[5] S. Lindsey, K. Sivalingam, C. S. Raghavendra, "Power Optimization in
Routing Protocols for Wireless and Mobile Networks," Handbook of
Wireless Networks and Mobile Computing, I. Stojmenovic, Ed., Wiley,
2001.
[6] C. E. Jones et al., "A Survey of Energy Efficient Network Protocols for
Wireless Networks," Wireless Network. Journal., vol.7, no. 4, Aug. 2001,
pp. 343-58.
[7] Peng-Jun Wan, Gruia Calinescu, Xiangyang Li, Ophir Frieder,
"Minimum-Energy Broadcast Routing in Static Ad Hoc Wireless
Networks", IEEE INFOCOM, Anchorage, AK, 2001.
[8] Laura Feeney, M. Nilsson, "Investigating the Energy Consumption of a
Wireless Network Interface in an Ad Hoc Networking Environment,"
IEEE INFOCOM, Anchorage, AK, 2001.
[9] David B. Johnson, David A.Maltz, Yih-Chun Hu, and Jorjeta G. Jetcheva,
"The Dynamic Source Routing Protocol for Mobile Ad Hoc Networks
DSR", Internet Draft, February 2002.
[10] P. Jacquet, P. Muhlethaler, T. Clausen, A. Laouiti, A. Qayyum, and L.
Viennot, "Optimized link state routing protocol for ad hoc networks," in
Proceedings of the 5th IEEE Multi Topic Conference (INMIC 2001).
[11] D. Kim, J.J. Garcia-Luna-Aceves, K. Obraczka, J. Cano, and P. Manzoni.
Power-Aware Routing Based on The Energy Drain Rate for Mobile Ad
Hoc Networks. In Proceedings of the IEEE International Conference on
Computer Communication and Networks, October 2002.
[12] J.-E. Garcia, A. Kallel, K. Kyamakya, K. Jobmann, J.-C. Cano, P.
Manzoni; "A Novel DSR-based Energy-efficient Routing Algorithm for
Mobile Ad-hoc Networks," in 58th IEEE Vehicular Technology
Conference (VTC Fall. 2003), 6-9 Oct. 2003, pp.: 2849 – 2854.
[13] C.Taddia, A.Giovanardi, G.Mazzini, M.Zorzi, “Energy efficient unicast
routing protocols over 802.11b,” in IEEE Global Telecommunications
Conference, (GLOBECOM’05), 28 Nov.-2 Dec. 2005.
[14] A.McCabe, A.Cullen, M.Fredin, L.Axelsson, “A power consumption
study of DSR and OLSR,” in IEEE Military Communications Conference
(MILCOM’05), 17-20 Oct. 2005, pp.1954 – 1960.
[15] K. Fall and K. Varadhan, (Eds.) "ns Notes and Documents", The VINT
Project. UC Berkeley, LBL, USC/ISI, and Xerox PARC, February 25,
2000. Available from http://www.isi.edu/~salehi/ns_doc/.