LIFETIME-AWARE MULTICAST ROUTING IN WIRELESS AD HOC
NETWORKS
Morteza Maleki and Massoud Pedram
Dept. of EE-Systems, University of Southern California, Los Angeles, CA 90089
{morteza, pedram}@usc.edu
Abstract. One of the main design constraints in mobile ad hoc
networks (MANETs) is that they are energy constrained. Hence,
routing algorithms must be developed to consider energy
consumption of the nodes in the network as a primary goal. In
MANETS, every node has to perform the functions of a router. So
if some nodes die early due to lack of energy and/or the network
becomes fragmented, then it will not be possible for other nodes in
the network to communicate with each other. This paper presents
a lifetime-aware multicast routing algorithm that maximizes the
ad hoc network lifetime by finding routing solutions that minimize
the variance of the remaining energies of the nodes in the
network. Extensive simulation results are provided to evaluate the
performance of the new routing algorithm compared to a number
of different metrics and in comparison to a variety of existing
multicast routing algorithms.
1 INTRODUCTION
An ad hoc network is one where in all nodes work independently
of any common centralized administrator. Each one of them
performs the tasks of a router. They should be self-adapting in that
if their connection topology changes, their routing tables should
reflect the change. Also, since they are mobile, they largely run on
finite energy sources (i.e., batteries.) This means that these nodes
are power-constrained. Hence, it is an important design constraint
for them to be power-aware or power-efficient. Furthermore, a
node should not be greedy about its reserving its own power
source since failure of some nodes in the network may result in
lack of connectivity among nodes that are alive.
The primary goal of the conventional multicast routing protocols
and algorithms has been to reduce the delay since most multicast
applications tend to be delay sensitive audio/video broadcasting.
Hence, most of the multicast routing protocols are designed to
construct a multicast tree that minimizes the communication
latency. Since the number of hops is a good heuristic metric for
capturing this latency, a multicast tree with the minimum number
of hops has been favored by most routing protocols ([1],[2],[3]).
We call this tree the Minimum Hop-count Tree (MHT). In
wireless ad hoc networks, there are two other criteria that make
routing design an even more complicated task, i.e., mobility and
power efficiency. The issue of mobility has extensively been
addressed in the literature. In fact, the performance of multicast
routing protocols has been evaluated in regard to their robustness
to link failure due to the mobility ([1],[3],[4],[5]). However, there
has been little work on developing a wireless multicast routing
protocol in which power is a key objective or constraint. More
precisely, although there have been some studies on the
construction of energy-efficient broadcast and multicast tree in ad
hoc networks ([6],[7]), most of these works require a global view
of the network and cannot be applied in a distributed way whereby
the nodes have only local knowledge.
This paper
1
addresses the problem of designing a lifetime-aware
multicast routing protocol and algorithm that can be applied to an
ad hoc network where nodes only have limited knowledge of the
network topology and power states of other nodes in the network.
The paper is structured as follows. Section 2 contains review of
prior work in energy-aware anycast (unicast, multicast,
and
broadcast) routing in ad hoc networks. Section 3 describes the
rationale and details of the proposed lifetime-aware multicast
routing algorithm and protocol. Section 4 describes the simulation
environment, the implementation, and the experimental results.
2 RELATED WORK
2.1 Energy-Aware Unicast Routing
Reference [8] proposes a routing algorithm based on minimizing
the amount of power (or energy per bit) required to get a packet
from a source to some destination. More precisely, the problem is
stated as:
(, )
P
i
j
ij
Min
π
π
∈
∑
(1)
where P
ij
denotes the power expended for transmitting (and
receiving) between two consecutive nodes, i and j (a.k.a. link
cost), in route π. This link cost can be defined for two cases: a)
when the transmit power is fixed; b) when the transmit power is
varied dynamically as a function of the distance between the
transmitter and intended receiver. For the first case, energy for
each operation (receive, transmit, broadcast, discard, etc.) on a
packet is given by [9]:
() _E packet b packet size c=× +
(2)
where b and c are operation-specific coefficients. Coefficient b
denotes the packet size-dependent energy consumption whereas c
is a fixed cost that accounts for acquiring the channel and for
MAC layer control negotiation.
The main disadvantage of the problem formulation of reference
[8] is that it always selects the least-power cost routes. As a result,
nodes along these least-power cost routes tend to “die” soon
because of the battery energy exhaustion. This is doubly harmful
since the nodes that die early are precisely the ones that are
needed most to maintain the network connectivity (and hence
useful service life). Therefore, it would be better to use a higher
power cost route if this routing solution avoids using nodes that
have a small amount of remaining battery energy. This
observation has given rise to a number of “battery-cost lifetime-
aware routing” algorithms ([10],[11]).
1
This research was sponsored in part by DARPA PAC/C program under
contract no. DAAB07-00-C-L516.
2.2 Minimum Energy Broadcasting
The main goal of minimum energy broadcasting is to reach from a
specific source to all the other nodes in the network in a multihop
transmission way with minimum total transmission energy
assuming that nodes have variable transmission power. In an ad
hoc network this may happen when we flood the network from a
specific source. Since the main use of flooding is in route
discovery it is important that flooding is done with minimum total
energy. Minimum energy broadcasting has been addressed as NP-
hard problem and there have been several works for finding
heuristics for this problem ([6]).
2.3 Energy-Aware Multicast Routing
The goal of energy-efficient multicast routing is to reach a subset
of nodes (one-to-many cast) that we will refer to as multicast
receivers from a multicast source, such that we have maximum
longevity of the paths between the source and the receivers. In
general, finding a minimum energy multicast tree is equal to
finding a minimum Steiner tree that is known to be an NP-hard
problem ([12]). Two related works on the energy-aware multicast
tree are as follows:
(1) Least-Cost shortest Path Tree (LPT): This is a tree obtained by
superimposing all least cost paths (or shortest paths) between
the source and each multicast receiver. The cost of each path
can be calculated by equation 1 or equation 3.
(2) Multicast Incremental Power Tree (MIPT) ([7],[13]): This tree
is obtained from the Broadcast Incremental Power (BIP) tree
proposed in [7]. The BIP algorithm consists of the following
steps. For all nodes i in the tree and all nodes j which are not in
the tree, we evaluate the incremental power cost of node i if
node j is connected to the tree through i. This cost is defined as
ρ’
ij
=ρ
ij
– ρ
i
, where ρ
ij
denotes the power level of node i if we
add node j to the tree through i while
ρ
i
denotes the current
power level of node i. Initially, the tree includes only the
source node (i.e., the broadcast initiator node.) A pair (i,j) that
results in the minimum value of ρ’
ij
is chosen and node j is
added to the tree through i. This procedure is repeated until all
nodes are included in the tree. The MIPT is generated by
pruning the broadcast (spanning) tree i.e. by eliminating all
sub-paths that are not required to reach the multicast receivers.
Reference [13] provides an improvement over [7] by dividing
ρ’
ij
by the remaining battery capacity of node i, thereby,
calculating as the cost function for link ij in the tree a quantity
that represents the normalized lifetime loss of node i if node j
is included in the tree through i. In [13], the final cost function
is obtained as the network lifetime loss to the power α, where
α is a fixed parameter (See Section 3.3 for a detailed
discussion of our proposed approach compared to [13].)
3 LIFETIME-AWARE MULTICAST ROUTING
3.1 Cost Function
The cost function that we adopt is the same as that is used in
Power-aware Source Routing (PSR) [14], which is an on-demand
source-initiated unicast routing algorithm that uses information
about the state of the charge in battery sources of nodes in the
network so as to maximize the network lifetime. More precisely,
PSR solves the following problem to find a unicast route
π
s
!
d
from source s to destination d at route discovery time t:
;,
()
( , ) ( )
(3)
where ( ) (1 ). (4)
()
:normalized transmit power of node i
:ful
i
sd i
i
i
i
i
i
i
iisd
sd
t
Ct Ct
Min
F
Ct
Rt
F
π
α
π
π
ρ
ρ
→
∈≠
→
=
=+
∑
l-charge battery capacity of node i
( ):remaining battery capacity of node i at time t
( ): a positive weighting factor that increases with
()
i
i
i
i
Rt
F
t
Rt
α
This cost function has two parts: one part captures the transmit
power level, whereas the other captures the remaining battery
capacity. The first term helps select a path with the minimum total
energy consumed, whereas the second term helps balance power
consumption over all nodes in the network as described next.
α
i
(t)
is inversely proportional to the ratio of the remaining capacity
over the full-charge capacity of the battery source. As this ratio
decreases and becomes less than a specified set of threshold
values one at a time,
α
i
(t) increases super linearly according to a
fixed schedule. In this way, nodes with very low remaining battery
capacity contribute a much higher value to the total path cost. In
other words, when a path from source to destination has some
nodes with very low residual battery, the cost of the path will be
very high, and therefore, PSR will behave similar to the Max-Min
battery cost routing. Note that
ρ
i
is normalized to the unicast
(and/or broadcast) reception cost of a node.
3.2 Neighbor Cost Effect in Multicast Routing
Assume that a multicast tree from the source to several receivers
has been constructed. The packet flow is coming out from source
and is terminated at leaves of the tree where the receivers are
located. We will refer to those intermediate nodes in the multicast
tree that have more than one child in the tree as multi-fanout
nodes (e.g., node A in Figure 1.) In ad hoc networks since the
MAC layer does not have the ability of multicasting ([9]), there
are two distinct methods to send out the packets from a multi-
fanout node:
(1) Multiple unicast: The parent node separately sends unicast
packets to every child node in the multicast tree,
(2) Single broadcast: The parent broadcasts the packets to all
nodes in its immediate neighborhood (which may include
nodes that are not in the multicast tree).
Reference [9] experimentally studied the power-optimal choice
between these two methods. According to its results, the multiple
unicast method results in much higher power consumption for the
sender (i.e., the parent node in the multicast tree.) Based on these
results, in our work, we adopt the single broadcast method at
multi-fanout nodes.
It should be noted that in sending broadcast packets as opposed to
sending unicast packets, there is no handshaking and
acknowledgement in the form of Request–To-Send/Clear-To-Send
(RTS/CTS) and ACK packets. Therefore, when using the single
broadcast method, all the nodes that are in the radio range of the
sender listen to the channel and receive the packet. As a result,
non-destined nodes will unnecessarily consume power to receive
the broadcast packet. In fact, these non-destined receiver nodes of
broadcast packets realize at the routing (or IP) layer that they are
not in the multicast tree that is identified in the IP header of
received packets. Therefore, it is difficult for the physical layer to
filter out non-destined packets at IP layer. This kind of filtering
has indeed not been implemented in current commercial WLAN
adaptors. As a result, multi-fanout nodes will find the single
broadcast method to be more beneficial to them from a power
dissipation viewpoint. One must, therefore, consider the power
consumption cost of all neighbors of nodes that broadcast packets
when calculating the cost of a multicast tree in which multi-fanout
nodes use a single broadcast method. This phenomenon, which we
will refer to as the neighbor cost effect, makes the problem of
finding a multicast tree with optimal cost quite complex.
Assuming that each node uses the cost function of equation (4),
the complete cost of a multicast tree, C(T,t), employing the single
broadcast method at time t can be written as follows:
()
;,
()
()
(,) .( )
()
() 2 ( ) 0
()
( ) : the degree of node in (including incoming and outgoing edges)
( ) : the set of
(1 )
(
5)
(
)
{
}
j
i
t
iTisi
jneighi jT
t
F
i
i
CTt
Rt
i
F
j
if i then else
Rt
j
iiT
neigh i
α
α
θ
θ
ρ
∈≠∉ϒ
∈∧∉
=+
≥
+
∑
∑
nodes that are in the radio range of node i
, :multicast source and set of multicast receiverss ϒ
As described earlier, the cost function of equation (4) includes a
power balancing part, which applies to all nodes that may
consume power. Because of the neighbor cost effect, some nodes
around the multicast tree consume power in their receiver parts.
Equation (5) adds to total cost of the tree cost of the nodes that are
not in the tree, but are in fact affected by the packet broadcasts at
the multi-fanout nodes.
Another issue concerning the single broadcast method at multi-
fanout nodes is that the farthest child from the parent determines
the broadcast transmission power of that transmitting node. For
example in Figure 1, the transmission power at node A is
Max(
ρ
1,
ρ
2). With considering the neighbor cost effect in multi-
fanout nodes, the multicast routing problem would even be more
challenging. Recall that finding a minimum energy-cost multicast
tree without considering the neighbor cost effect is equivalent to
that of finding a minimum Steiner tree which is NP-hard. As a
result the problem of finding a lifetime-aware multicast tree with
consideration of the neighbor cost effect is also an NP-hard
problem.
There are many algorithms for finding a tree with near optimal
cost ([15],[16]). Although it is possible to modify some of these
algorithms to account for the neighbor cost effect at multi-fanout
nodes, this approach is ill-advised in our context because these
algorithms are too complex and require global information about
the network connectivity graph in order to be applied. However,
we are interested in finding solutions that can be deployed in an ad
hoc network where nodes only have local knowledge about
themselves and perhaps their neighboring nodes and must do the
route discovery in a distributed, ad hoc manner (no global
depository of information exists.) Furthermore, in ad hoc
networks, the underlying network topology (connectivity graph)
changes dynamically due to the mobility and link failure. Hence,
ad hoc routing algorithms should be able to periodically update
their routes. The routing update cost should be rather low.
Figure 1: Neighbor cost effect in wireless networks.
3.3 Algorithm Design for Constructing a Lifetime-aware
Multicast Tree
In this section, we present an algorithm for constructing a
Lifetime-aware Multicast Tree (LMT) for multicast routing in ad
hoc networks (with single broadcast at multi-fanout nodes of the
tree.) We will show that this algorithm can be deployed in ad hoc
environments with some flooding overhead. Flow of the LMT
algorithm is as follows:
(1) Using the cost function given by equations (3) and (4), find
least-cost path between the source and each receiver in the
multicast tree
(2) Sort receivers in order of increasing path cost to the source
(3) Include the least-cost path from the source to the first receiver
in the sorted list as part of the LMT
(4) Extend the LMT by finding the least-cost path from the
existing LMT to the next receiver in the sorted list
(5) Repeat step 4 until all receivers are connected to LMT
The rationale for connecting receivers with lower path costs to the
source earlier than those with higher path costs (step 2) is that the
paths included earlier in the LMT tend to carry (relay) more data
traffic (because of step 4 in which subsequent receivers may
connect to some intermediate node in the partially constructed
LMT in order to establish their connection to the source). Step 4
is the key step in the LMT algorithm and is explained below.
The cost of a path, C(
π
r
,t), from a multicast receiver, r, to a
partially constructed multicast tree, T, at contact point j and at
time instance t can be calculated as follows:
()
;
()
()
(,) (1).()
()
( )
()
( , ) ( , )
(
6)
t
i
iir
rj
t
k
kneighj kT
rj
F
i
Ct
i
rj
Rt
i
F
k
Rt
k
CtMinC t
r
jT
α
π
α
πρ
ππ
→
∈≠
→
∈∧∉
→
∑
=+ +
∑
=
∈
(7)
R
S
R
R
R
ρ1 ρ2
Broadcast with power
Max(ρ1, ρ2)
A
Affected
area
Multicast traffic
flow
R
S
R
R
R
ρ1 ρ2
Broadcast with power
Max(ρ1, ρ2)
A
Affected
area
Multicast traffic
flow
In equation (6),
π
r
!
j
denotes the set of nodes from receiver r
(exclusive) to contact point j (inclusive). neigh(j) consists of all
nodes that are in the transmit range of node j which is determined
by the farthest child of node j (which is on
π
r
!
j
). The normalized
transmit power for node j (the contact point) is recalculated to be
included in equation (6) if node j has to increase its power level to
connect receiver r through path
π
r
!
j
to the tree.
Notice that compared to [13], our proposed multicast tree
construction algorithm is more efficient and can be easily
deployed in a lifetime-aware multicast routing protocol (see
below.) More precisely, at each step in the LMT algorithm, we
only calculate the least cost path from the set of unconnected
multicast receivers to the partially constructed tree T. In contrast,
at each step of the MIPT algorithm, we ought to calculate the cost
of all links between nodes already included in T and those that are
not included in order to identify the lowest cost link. These
modifications in tree construction results in a significant
performance improvement for the LPT over MIPT when the
number of multicast receivers is small compared to the total
number of nodes in the ad hoc network. In addition, [13] uses a
fixed value for
α
and does not consider the neighbor cost effect.
3.4 Protocol Design for Constructing a Lifetime-aware
Multicast Tree
In this part we propose a high-level routing protocol for LMT. We
have selected an on-demand approach that has already been
deployed in ODMRP ([1]). In ODMRP group membership and
multicast routes are established and updated by the source. While
a multicast source has data to send, it floods the networks with
Join-Request (JREQ) packets. Flooding is like asking every node
in the network to find a group of multicast receivers. When a node
receives a non-duplicate JREQ, it rebroadcasts it in the network.
When a multicast receiver receives the JREQ packet it replies
back to the source. The result of flooding and reply-back
procedure is that a muticast tree rooted at the source is
constructed. However, the metric for constructing this tree in [1] is
the number of hops. In other words, this is a least-cost path tree
with respect to the number of hops between the source and each of
the multicast receivers. The reason is that during the flooding
process, every node simply rebroadcasts only the first arriving
JREQ packet and drops any other copies of the packet that arrive
later and from other paths. In addition, receivers immediately
reply back to the first arriving JREQ packet. The first arriving
JREQ has generally traversed the path that has least number of
hops because that path usually has the lowest delay. In ODMRP
the source periodically floods the network to refresh the
membership information and update the routes and thereby the
muticast tree.
The process of finding the least-cost path tree is similar to that
used in ODMRP except that we use cost function given in
equations (2) and (3). Furthermore, during the flooding process,
every node may pass on several copies of the JREQ packet from a
multicast source. In particular, every node passes the first arriving
JREQ and then turns on a timer whose count is proportional to the
number of hops in the first arriving JREQ. All JREQ’s will be
passed through this node until the timer expires after which the
node blocks subsequent JREQ’s from the multicast source. As
long as the node timer is not expired, if the node receives a new
copy of a JREQ packet, it examines the cost in the header of
JREQ packet. If that cost is less than the cost of a previous copy
of that JREQ packet that has already passed through the node,
then it will pass on the new copy as well; otherwise, it will drop it.
Furthermore, when a node passes on a JREQ, it adds its own cost
to the header of the packet so that the cost of a packet is updated
as it traverses along a path. Muticast receivers also have their local
timers and reply back on (i.e., select) the path that has the least-
cost when their timers expire. To implement the LMT algorithm
in an ad hoc network, a number of steps must be followed as
described below:
(1) The source of multicast tree floods the network to find a least-
cost path tree to all multicast receivers. This process has
already been explained. One more note is that the multicast
receivers add the cost of the path that they select to the header
of the reply-back packet before sending this packet to the
source.
(2) The source sorts the receivers in increasing order of their
respective path costs and sends a path confirmation packet to
the receiver whose path has the least cost. Next,
it removes
that receiver from the sorted list and initializes the multicast
tree, T, to consist of the least-cost path from the source to that
receiver.
(3) The source sends a FLOOD command to the next receiver in
the sorted list and asks this receiver to flood the network and
search for any of the nodes in T.
(4) The receiver that receives a FLOOD command from the
source, starts flooding the network searching for the nodes in
T. This is equivalent to starting another source initiated
flooding in multicast routing where this flooding source is the
receiver under considerations and the receivers are all the
nodes in T.
(5) Every node that is in T replies back to the flooding receiver
along the least-cost path from that node to the flooding
receiver while accounting for the neighbor cost. Note that
nodes can easily obtain neighbor information, including
number of their neighbors, from the MAC layer.
(6) The flooding receiver chooses the path with least cost to some
node in T. It then sends information about the subpath that
must be added to T, to the multicast source.
(7) The multicast source updates T to include the new subpath to
the flooding receiver, removes that receiver from the sorted list
and repeats steps 3 through 6 until all receivers in the multicast
group are connected to T (the list becomes empty.)
(8) The proposed LMT protocol optimizes the transmit power
level of each node of the multicast tree after T has been
constructed in the way that each node in the LMT will
determine its own minimum transmit power level based on the
distance of its farthest child node in the tree.
Clearly the LMT protocol causes an increase in the number of
flooding procedures, which is proportional to the number of
receivers in the multicast group (cf. step 4 above.) However, the
protocol works particularly well for a scenario with the “hard
state” method as follows. Initially a small set of multicast
receivers exists and the source forms an LMT tree according to
the abovementioned protocol. When a new multicast receiver
arrives, it should join the tree by following steps (4) through (6)
except that the source does not need to send any FLOOD
command to the new receiver as described in step (4) and the new
receiver starts flooding the network to find a node as the contact
point to the tree. Recall that in the “hard state” join/leave method,
there is no periodic refresh to update the multicast tree. Instead
nodes send explicit commands to join or leave the tree whenever
they want to do so. Furthermore, with in the “soft state” join/leave
method, if the number of initial receivers is large, then as the size
of the multicast group increases, the network lifetime gain of the
LMT increases (cf. Section 4). Finally, flooding is done by
sending small control packets. There are ways to reduce the
overhead of flooding process ([17],[18]).Therefore, the cost of the
flooding processes can be made small.
4 EXPERIMENTAL RESULTS
4.1 Simulation Setup
To simulate our routing algorithms, we developed an event-driven
simulator (called LRSim for Lifetime-aware Routing Simulator)
in C++. The block diagram of the simulator is shown in Figure 2.
LRSim is a high-level simulation environment that can implement
any kind of routing algorithm without implementing the MAC
layer. Nodes are randomly distributed in an area and their
locations are given as input to LRSim. For our simulations, we
used 50 nodes, which are uniformly located in a 1000 by 1000 m
2
area. The nodes were randomly chosen as a source or a receiver
member of a multicast connection. 100 multicast connections with
random duration and random starting times were established
during the simulation time, which is 20,000 secs. Connections had
CBR (Constant Bit Rate) traffic with the rate of 3 packets per sec.
Each node was randomly assigned an initial energy, which varied
between 1500, and 3000 units of energy. The nodes have four
different transmission power levels for four radio ranges: 150,
125, 75, and 50 meters. The energy cost for transmitting and
receiving a packet are as equation (2) where the coefficients (a
and b) are taken from [9] which were experimentally measured.
The coefficients for different states (and for transmit range of 150)
are as follows:
Unicast send
1.9 x
p
acket_size+454
(µW.sec)
Broadcast send
1
.9 x packet_size+266
(µW.sec)
Unicast receive
0
.5 x packet_size+356
(µW.sec)
Broadcast receive
0.5 x packet_size+56 (µW.sec)
In transmit mode, value of a varies as a function of transmission
power level or range ([20]) and accordingly the above coefficients
are calculated for other transmission ranges (125, 75 and 50
meters.) In each simulation run, some routing algorithms are
simulated and finally the results of different simulation runs are
analyzed and compared. After each packet generation event, the
remaining energies of the nodes participating in data transmission
are decremented to reflect the cost of transmitting, relaying or
receiving the packet. When the remaining energy of a node
reaches zero that node is considered dead for the rest of the
simulation and does not participate in any subsequent
transmission or reception.
Figure 2: LRSim flow diagram.
4.2 Simulation Results
We have done several simulations and have measured the effect of
a number of different multicast routing algorithms on the network
lifetime, the RMS value of remaining energy, the packet delivery
ratio, and energy consumption per transmitted packet. The
performance of the minimum hop-count tree (MHT), the multicast
incremental power tree (MIPT), the least-cost path tree (LPT), and
the lifetime-aware multicast tree (LMT) have been evaluated and
compared with respect to the abovementioned metrics. We have
selected for our experimental results a variant of the MIPT
algorithm as proposed in reference [13], where the MIPT cost
function is proportional to the normalized lifetime loss. For a fair
comparison of MIPT and LMT, the value of α that has been
chosen for MIPT is equal to the average value of α(t) during the
lifetime of each node.
We have applied an on-demand approach, which periodically
refreshes the multicast tree. The timing period of refresh in our
simulation was set to 50 sec. In terms of the cost function, MHT
uses the hop-count metric whereas MIPT, LPT and LMT rely on
the cost function given in equation (4). We can state that the
performance of MHT is the same as that of the ODMRP.
We may define the network lifetime as the total elapsed time from
the state of full battery charge for all nodes in the network to a
state in which a fixed number (or percentage) of the nodes in the
network die due to energy source exhaustion. Figure 3 shows the
ratio of a number of nodes that are alive to the total number of
nodes during part of simulation time at different time instances
(when multicast group size is 15.) As can be seen, in MHT nodes
start dying out sooner but with a rather uniform rate. In LMT,
LPT and MIPT, the nodes start dying later but die more rapidly. In
terms of the network lifetime as will be shown, LPT performs
better than MHT but worse than MIPT and LMT for large group
size. Although the initial multicast group size was set to 15 in this
Moving path
model
Node
generator
Number of nodes
Positions
Transmission
Range
Multicast
call
generator
Number of
Calls
Call Duration
S
et of Sources and
Receivers
Packet rate
Routing
Algorithms
Event handler
and Scheduler
Mobility
manger
Speed
Result
model
Node
generator
Positions
Initial Energy
Multicast
call
generator
Number of
Calls
Call Duration
S
et of Sources and
Receivers
Packet rate
Routing
Algorithms
and Scheduler
manger
Speed
Result
MHT
LMT
MIPT
Moving path
model
Node
generator
Positions
Transmission
Range
Multicast
call
generator
Number of
Calls
Call Duration
S
et of Sources and
Receivers
Packet rate
Routing
Algorithms
Event handler
and Scheduler
Mobility
manger
Speed
Result
LPT
model
Node
generator
Positions
Space
Boundary
Initial Energy
Multicast
call
generator
Number of
Calls
Call Duration
S
et of Sources and
Receivers
Packet rate
Routing
Algorithms
and Scheduler
manger
Speed
Result
MIPT
