Research Article
Throughput Capacity of Opportunistic Routing in Wireless
Sensor Networks
Recently, the idea of opportunistic routing (OR) has been widely explored to cope with the unreliable transmissions by exploiting
the broadcast nature and spatial diversity of the wireless medium in order to improve the performance of wireless sensor networks.
However, there are few theoretical analyses on the maximum throughput of OR WSNs. This paper is the first attempt to conduct a
theoretical analysis on aggregate throughput capacity of OR in multihop many-to-one WSNs with consideration for lossy link and
transmission fairness. By capturing the key characteristics of forwarding candidate set in OR networks, we propose the cumulative
delivery transmission model. Then we introduce the concept of concurrent schedulable set to represent the constraints imposed
by the transmission conflicts of OR, and formulate the optimal aggregate throughput problem as a maximum concurrent flow
linear programming problem. Simulation results demonstrate that the OR design derived from our analysis model often yields
noticeably better throughput than traditional unicast routing protocols and the OR design derived from existing analysis model
under a range of scenarios.
ni1
the packet only when it is of the highest priority among the
nodes which have received the packet. For example, if the
Pii1
Pi2i1 Pi1d first node ni1 in the set has received the packet successfully,
ni Pii2
it forwards the packet towards the destination while all
ni2 Pi3i1
nd other nodes suppress duplicate forwarding. Otherwise, the
Pii3 Pi2i3 second node ni2 , the third node ni3 , the forth node ni4 , and
Pi3d
so forth. When no forwarding candidate has successfully
Pii4 ni3 received the packet, the sender ni will retransmit the packet
Pi2i4
Pi4d
if retransmission is enabled. The forwarding reiterates until
Pi3i4
the packet is delivered to the destination node nd .
ni4
2.2. Packet Level Lossy Wireless Interference Model. Wireless
Figure 1: Opportunistic routing in the subgraph of Roofnet. interference is a key issue affecting throughput [14]. As OR
exploits the broadcast nature and spatial diversity of the
wireless medium by selecting unreliable wireless links in
services. Furthermore, the derivation of throughput of OR order to achieve better performance, the wireless interference
may suggest novel and efficient candidate selection and model should be nonbinary and reflect the features of
prioritization schemes. lossy links. Therefore, we propose the packet level lossy
We view the cumulative delivery transmission model wireless interference model to define the conditions for a lossy
and the aggregate throughput capacity analysis framework wireless transmission. This model is similar to the receiver
for OR WSNs with consideration for lossy links and interference model of packet level loss rates introduced in
transmission fairness as the key contributions of our work. [14].
Simulation results show that the OR design derived from our The packet level lossy wireless interference model takes
analysis model often achieves noticeably better throughput as input traffic demand and received signal strength (RSS)
than traditional unicast routing protocols and the OR design between pairs of nodes. It then estimates the rate at which
derived from existing analysis model under a range of each sender will transmit and the rate at which each receiver
scenarios with different traffic demands. will successfully receive the packets. To compute the link
The rest of the paper is organized as follows. Section 2 packet delivery ratio, we model the inherent medium loss
introduces the system model. We present details of cumu- as well as collision loss. In this paper, we assume that the
lative transmission delivery model and the framework of packet transmissions among individual nodes can be finely
computing the optimal network throughput bounds of OR controlled and carefully scheduled by an omniscient and
in Section 3. Simulation results are presented and analyzed omnipotent central entity, which guarantees that there is
in Section 4. In Section 5, we discuss related works. Section 6 no MAC contention. So different from [14], we only take
concludes the paper. as the asynchronous loss that occurs when at least one
sender cannot carrier sense the other because there is no
synchronous collision. The link packet delivery ratio between
2. System Model node nm and node nn initiated by node nsi is
2.1. Opportunistic Routing Framework. The basic module of
si COL
Pm,n = 1 − 1 − LRSS
mn × 1 − Lmn|si , (1)
opportunistic routing employed by the well-known Roofnet
[13] is shown in Figure 1. Assume node ni is transmitting
a packet to the destination node nd , and the other four where LRSS
mn is the inherent packet loss rate on link lmn which is
nodes are the candidates for forwarding. Each link packet a constant value for node nsi ; LCOL
mn|si is asynchronous collision
delivery ratio (LPDR), denoted as Pk, j for the link between packet loss on link lmn caused by node nsi , which can be
node nk and node n j , is shown in Figure 1. We denote the modeled as
set of nodes which receive the packet from node ni as the 8 f
neighboring node set Ci of node ni . Each node selects a LCOL b
mn|Si = 1 − 1 − Lmn|Si , (2)
subset of its neighbors as forwarding candidate set (FCS)
and assigns a priority to each of the candidates, denoted as where f is the length of the packet transmitted by node nm ,
Fi := ni1 , ni2 , ni3 , ni4 for node ni shown in Figure 1, which Lbmn|Si is the collision bit error rate defined as
includes all the nodes selected to be involved in the local
opportunistic forwarding based on a particular selection 1
strategy. Note that, Fi is an ordered set, in which the order Lbmn|Si = × expγmn|Si /2 , (3)
2
of the elements corresponds to their priority in forwarding a
received packet. where γmn|Si = Rmn /(An + Rsi n ) is the signal to interference
When node ni broadcasts a packet, the nodes which plus noise ratio (SINR) of link lmn interfered by node nsi , Rmn
indeed receive the packet in its forwarding candidate set Fi is the received signal strength at node nn , An is the ambient
respond. One of the candidate nodes continues the forward- noise of node nn , and Rsin is the interference at node nn from
ing based on their relay priority –A candidate will forward node nsi .
International Journal of Distributed Sensor Networks 3
Inferred from (1), (2), and (3), we can model the link
ni
packet delivery ratio of lmn over node nsi as a function of Rmn {ni1 , ni2 , ni3 , ni4 }
and Rsin :
Ri,i2 Ri,i1
ni2
8 f
si 1 {ni1 , ni3 , ni4 , nd } Ri1,i2 ni1
Pm,n = 1 − 1 − LRSS
mn × 1 − × expRmn /2(An +Rsi n ) . (4)
2 {nd }
Ri,i4 Ri,i3
Ri2,i4 Ri2,i3
Similarly, we can get the cumulative link packet delivery Ri1,i3
ratio (CPR), denoted as Pm,n , as the following:
ni4 ni3
8 f {nd } Ri3,i4 {ni1 , ni4 , nd }
1
Pm,n = 1 − 1 − LRSS
mn × 1 − × expRmn /2(An + k =/ m Rkn ) .
2
(5) Figure 2: Weighted interference graph for OR Roofnet.
We say that a directed link li j from node ni to n j is usable usable. Then the cumulative transmission delivery rate of
when Pm,n is larger than a positive threshold Ptd . Notice that node ni in that particular time is defined as:
our definition of Ptd in OR networks which is generally less
than 20% is much smaller than that in traditional routing
r −1
k
networks (e.g., the Ptd in [15] is more than 80%). CDRi = ϕk · pi,ik · 1 − ϕt · pi,it , (6)
k=1 t =0
joint effect and asymmetry of the transmission interference communications between source node set Ns {ns1 , ns2 . . . nsi }
in OR networks. Node m is termed to have a “transmission and a single destination nd , we formulate the optimal
conflict” caused by some node set Cm when node m and node aggregate throughput problem of OR WSNs as a linear
set Cm are simultaneously transmitting and the cumulative programming corresponding to a maximum concurrent flow
delivery rate of node m is less than a certain threshold problem under additional constraints, in (7)–(16) which
CDRthresh (i.e., CDRm ≤ CDRthresh ) due to the joint introduce the LP formulations used to optimize the aggregate
interference of the nodes in Cm on its forwarder candidates. throughput for Many-to-One OR WSNs,
The node set Cm is named as the cooperative conflict set over
node m (denoted by CCSm ), and i-CCSm denotes that there
are i nodes in the cooperative conflict set, which can be Max feffsi ,si
nsi ∈Ns
obtained via the weighted interference graph.
s j ,s j (7)
s.t. γ × feff ≤ feffsi ,si ≤ Ssi
3.3. Cumulative Delivery-Based Concurrent Schedulable Sets.
∀nsi ∈ NS , ∀ns j ∈ Ns \ nsi ,
We introduce the notion of cumulative delivery based
d,s s ,s
concurrent schedulable set (CSS) in the packet level lossy feff i = feffi i , (8)
wireless interference model for OR WSNs, which is the nsi ∈Ns nsi ∈Ns
foundation of our method of computing optimal aggregate
throughput. A CSS is defined as a set of nodes, in which when feffk,si ≤ i,si
feff
they are transmitting simultaneously, none of which has a ni ∈Fk (9)
transmission conflict caused by all the other nodes in that ∀nsi ∈ Ns , ∀nk ∈ Nc \ {nd },
set. If adding any one more node into a CSS will result in
the usability status of any nodes in that set, the CSS is called feffk,si ≤ fact
k,si
× CDRk
(10)
maximal CSS. (In the following of this paper, CSS refers ∀nsi ∈ Ns , ∀nk ∈ Nc \ {nd },
to maximal CSS.) The idea is to allow a node to transmit d,si
as long as it can deliver some throughput which meets the fact =0 ∀nsi ∈ Ns , (11)
requirement of cumulative delivery rate threshold to some of i,sk
fact ≤ capi ∀ni ∈ Nc \ {nd }, (12)
its forwarder candidates. The maximal CSSs can be obtained nsk ∈Ns
through Algorithm 1.
T
j,s
In (7)–(16), fact i denotes the amount of data flow derived guarantees a high throughput by balancing the tradeoff
j,s between packet loss rate and transmission rate. While in [12]
from source node nsi actually transmitted by node n j ; feff i
denotes the amount of data flow derived from source node it schedules the transmissions based on node set without any
nsi required to be transmitted by node n j ; Nc denotes the delivery rate consideration which will induce the conflict
whole node set in the interference graph G; Ssi is the data between packet loss rate and transmission rate. That is why
generation rate of the source nsi ; Fk (i) denotes the ith forward our model achieves higher throughput than that of [12]. The
candidate in node nk ’s forwarder candidate set Fk ; Fk−1 (ni ) detail comparison will be given in the next section.
denote the index of node n j in node nk ’s forwarder candidate
set Fk ; Capi denotes the transmission capacity of node ni ; P j,k 3.5. Hardness Result. Now, we present a hardness result
denotes the cumulative link packet delivery ratio between for computing the optimal aggregate throughput with LP
node n j and node nk ; ϕmj,k denotes the indicator function on formulation shown in (7)–(16). The part of flow conser-
forwarder candidate nk of node n j when n j is in a particular vation and fairness constraints (i.e., constraints (7)–(12))
concurrent schedulable set CSSm ; ϕmj,k = 1 indicating nk is is a simple structure on which a linear objective function
usable, and ϕmj,k = 0 indicating otherwise. can easily be optimized. While for the part of wireless
The maximization states that we wish to maximize the interference constraints (i.e., constraint (13)–(16)), the con-
sum of flow out of the all source nodes, in which con- current schedulable set, on the other hand, is a difficult
straints (7)–(12) denote the flow conservation and fairness structure and no characterization of it is known (because
constraints and constraints (13)–(16) denote the wireless there may be exponentially many CSSs [17]). When each
interference constraints. The constraint (7) represents the node has only one forwarding candidate, OR degenerates to
the traditional routing. Therefore, finding all the concurrent
flow validity and fairness, that is, the amount of effective
schedulable sets is at least as hard as the NP-hard problem of
flow derived from each source node is constrained by its
finding the independent sets in [15] for traditional routing.
data generation rate and the amount of effective flow derived
Correspondingly, it is NP-hard to find the optimal aggregate
from other source node. The constraint (8) states that flow-
throughput in OR WSNs.
conservation, that is, the amount of outgoing flows from all
Even though it is a NP-hard problem, complexity can
source nodes is equal to the amount of incoming flow of
be reduced by taking into consideration that transmission
the sink node. The constraint (9) indicates that the amount
conflicts always happen for nodes within certain range.
of effective flow of every node, except the sink node, is
We can look at heuristics for obtaining bounds on the
constrained by the total amount of effective flows of those
throughput to approximate the optimal aggregate of OR
nodes whose forwarder candidate sets involve that node.
WSNs.
The constraint (10) denotes that the amount of effective
flow of each node, except sink node, is constrained by the
amount of outgoing flow of that node and its cumulative 3.6. Lower Bound. The problem of deriving a lower bound
delivery rate. The constraint (11) says that the outgoing flow is equivalent to the problem of finding a network aggregate
from sink node is 0. The constraint (12) represents that throughput that has a feasible schedule to achieve it. Suppose
the amount of outgoing flow on each node cannot exceed we can get T CSSs through Algorithm 1. If we replace the
its transmission capacity. The constraint (13) states that at schedule restrictions imposed by all the CSSs in the original
any time, at most one CSS will be scheduled to transmit. linear program (i.e., constraints (13)–(16)) with those
The constraint (14) restricts that the scheduled time fraction imposed by the found T CSSs, the resulting throughput
should be non-negative. The constraint (15) indicates that always has a feasible scheme, and therefore constitutes a
the amount of effective flow delivered by each node, except lower bound on the maximum achievable throughput. To
source nodes, is constrained by the total amount of flow tighten the bound, we should consider using maximal CSSs
that can be delivered in all activity periods of those nodes as much as possible.
whose effective forwarder candidate sets involve that node.
The constraint (16) denotes that the amount of outgoing 3.7. Upper Bound. To derive an upper bound, we consider a
flow on every node, except sink node, is constrained by the maximal set of nodes in the interference graph that mutually
sum of the activity periods of the concurrent schedulable sets have transmission conflicts with each other which is termed
which it is a member. as one-node level entirely cooperative conflict set (ECCS1 ). The
The key difference of our maximum of flow formulations total utilization of the nodes in each ECCS1 is less than 1.
from the formulations for opportunistic routing in [12] Similarly, the i-nodes level entirely cooperative conflict set
lies in the methodology we use to schedule concurrent (ECCSi ), a maximal set of nodes that any node can transmit
transmissions. In our proposed model, we involve the factor packet simultaneously with part of other nodes in the set but
of the cumulative delivery ratio in flow conservation and has a transmission conflict caused by all the other i nodes. In
fairness considerations which can accurately captures OR’s ECCSi , the total utilization of the nodes is less than i-1, the
capability of delivering throughput opportunistically. With proof for which is not included due to space constraints.
the construction of cumulative delivery-based concurrent We define i-ECCS as denotes a set of ECCSs including
schedulable sets, we are able to schedule the transmissions ECCS1 , ECCS2 , . . . and ECCSi , which gives us a constraint
based on node set whose cumulative delivery ratio exceeds on the usage vector. For example, we can find all the X
a threshold the same as in traditional routing, which ECCS1 s in 1-ECCS and write corresponding constraints to
6 International Journal of Distributed Sensor Networks
0.7
S21 n22 n23 n24 n25 n26 n27
0.6
Aggregate throughput
0.5
S31 n32 n33 n34 n35 n36 n37
0.4
0.3
S41 n42 n43 n44 n45 n46 D47
0.2
0
1 2 3 4 5 6 7 8 9 10
S61 n62 n63 n64 n65 n66 n67 Number of acquired CSSs ×104
Figure 3: 7×7 grid WSN. Figure 4: Aggregate throughput bounds of COR with varying
constraints.
define a polytope. We can then maximize the aggregate 4.1. Simulation Setup. The simulated network has 49 sta-
throughput over the intersection of this polytope with tionary nodes uniformly distributed in a 900 m × 900 m
the flow constraints in the original linear program (i.e., square region shown in Figure 3. We fix the seven nodes
constraints (7)–(12)). Thus an upper bound can be obtained. in the first column of the grid as the source candidates,
and all of them communicate with the same destination
λi ≤ 1 1 ≤ m ≤ X, node located in the center of the last column. The PRR
(17) threshold ptd in multipath routing and OR routing is set
ni ∈ECCS1m
to 0.8 and 0.1, respectively. We assume that the PRR is
λi ≥ 0 ∀ni ∈ ECCS1m , 1 ≤ m ≤ X, (18) inversely proportional to the distance with random Gaussian
deviation of 0.1 and that all the wireless links have an
i,sk
fact ≤ λi Capi ∀ni ∈ Nc \ {nd }, 1 ≤ m ≤ X. identical capacity (i.e., speed) of 1 unit. Similar to [18],
(19)
nsk ∈NS we use a single interference range 500 m for simplicity. The
performance metric is the aggregate throughput.
These constraints may result in a loose bound since there
may not be abundant ECCS1 s. To tighten the upper bound, 4.2. Aggregate Throughput Bounds of Traditional Multipath
we further augment the LP formulation with the more ECCS Routing and OR Under COR Model. In this section we
constraints by larger i-ECCSs. consider a scenario in the 7 × 7 simulated network that the
Unfortunately, it is computationally expensive to find all two nodes S31 , S51 are selected as the sources and the sending
levels of the ECCSs, and even if we could find them all, there rate of each source as 1.
is still no guarantee that our upper bound will be tight. Figure 4 shows the upper and lower bounds of aggregate
throughput for OR WSNs derived from our COR model with
4. Performance Evaluation varying constraints. We can see that the upper bound quickly
converges to the stable value. More specifically, the more
This section uses Matlab to investigate the performance of ECCS constraints being employed, the more accurate upper
OR variant: ExOR [2] in many-to-one WSNs. The section bound approaching the optimal value is achieved. The lower
is organized as follows. We first present the illustrative bound, on the other hand, steadily converges to the optimal
results that demonstrate the flexibility of our cumulative value as the number of acquired CSSs increases. However, the
delivery-based OR analytical model (COR) to discuss the upper bound and lower bound cannot converge together due
issue of convergence of the upper and lower bounds on to the fact that the ECCS constraints alone are not sufficient
the optimal aggregate throughput. We also investigate the to guarantee optimality of the upper bound, even all ECCS
maximum aggregate throughput in a wide range of scenarios constraints are involved.
with different traffic demands and compare them with those We pick N relaying nodes from 46 available nodes (i.e.,
derived from of the traditional multipath routing analytical they do not include the two sources and the destination),
model (MPR) [15] and the individual delivery-based OR at random, and without replacement. Then, in such a 2-
routing analytical model (IOR) [12]. flow network, the resulting normalized maximum aggregate
International Journal of Distributed Sensor Networks 7
Per-flow sending rate = 1 throughput derived from both OR models increases as the
60
richer connectivity provided by additional nodes allows for
Normalized maximum aggregate throughput
80
0.65
70
0.6
60
0.55
50
0.5
40
0.45
30
0.4
0.35 20
0.3 10
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Per-flow sending rate Per-flow sending rate
Conference on Applications, Technologies, Architectures, and [16] H. Zhai and Y. Fang, “Impact of routing metrics on path
Protocols for Computer Communications (SIGCOMM ’05), pp. capacity in multirate and multihop wireless ad hoc networks,”
133–144, Philadelphia, Pa, USA, October 2005. in Proceedings of the 14th IEEE International Conference on
[3] M. Kurth, A. Zubow, and J.-P. Redlich, “Cooperative oppor- Network Protocols (ICNP ’06), pp. 86–95, November 2006.
tunistic routing using transmit diversity in wireless mesh [17] E. L. Lawler, J. K. Lenstra, and A. R. Kan, “Generating all
networks,” in Proceedings of the 27th IEEE Conference on maximal independent sets: NP-hardness and polynomial time
Computer Communications (INFOCOM ’08), pp. 1984–1992, algorithms,” SIAM Journal on Computing, vol. 9, no. 3, pp.
Phoenix, Ariz, USA, April 2008. 558–565, 1990.
[4] Z. Hu and C.-K. Tham, “HOF: hybrid opportunistic forward- [18] H. Zhai and Y. Fang, “Physical carrier sensing and spatial
ing for multi-hop wireless mesh networks,” in Proceedings of reuse in multirate and multihop wireless ad hoc networks,”
the IEEE International Conference on Communications (ICC in Proceedings of the 25th IEEE International Conference
’08), pp. 2963–2968, May 2008. on Computer Communications (INFOCOM ’06), pp. 1–12,
[5] B. Radunović, C. Gkantsidis, P. Key, and P. Rodriguez, “An Barcelona, Spain, April 2006.
optimization framework for opportunistic multipath routing [19] Y. Li, L. Qiu, Y. Zhang, R. Mahajan, and E. Rozner, “Pre-
in wireless mesh networks,” in Proceedings of the 27th IEEE dictable performance optimization for wireless networks,” in
Conference on Computer Communications (INFOCOM ’08), Proceedings of the ACM Conference on Applications, Technolo-
pp. 241–245, Phoenix, Ariz, USA, April 2008. gies, Architectures, and Protocols for Computer Communications
[6] Z. Zhong and S. Nelakuditi, “On the efficacy of opportunistic (SIGCOMM ’08), pp. 413–425, Seattle, Wash, USA, August
routing,” in Proceedings of the 4th Annual IEEE Commu- 2008.
nications Society Conference on Sensor, Mesh and Ad Hoc
Communications and Networks (SECON ’07), pp. 441–450,
June 2007.
[7] C.-P. Luk, W.-C. Lau, and O.-C. Yue, “An analysis of oppor-
tunistic routing in wireless mesh network,” in Proceedings of
the IEEE International Conference on Communications (ICC
’08), pp. 2877–2883, May 2008.
[8] J. Wu, M. Lu, and F. Li, “Utility-based opportunistic routing
in multi-hop wireless networks,” in Proceedings of the 28th
International Conference on Distributed Computing Systems
(ICDCS ’08), pp. 470–477, July 2008.
[9] S. Chachulski, M. Jennings, S. Katti, and D. Katabi, “Trading
structure for randomness in wireless opportunistic routing,”
in Proceedings of the ACM Conference on Applications, Tech-
nologies, Architectures, and Protocols for Computer Communi-
cations (SIGCOMM ’07), pp. 169–180, Kyoto, Japan, August
2007.
[10] S. Katti, D. Katabi, H. Balakrishnan, and M. Medard, “Symbol-
level network coding for wireless mesh networks,” in Pro-
ceedings of the ACM Conference on Applications, Technologies,
Architectures, and Protocols for Computer Communications
(SIGCOMM ’08), pp. 401–412, Seattle, Wash, USA, August
2008.
[11] R. C. Shah, S. Wiethölter, and A. Wolisz, “Modeling and
analysis of opportunistic routing in low traffic scenarios,” in
Proceedings of the 3rd International Symposium on Modeling
and Optimization in Mobile, Ad Hoc, and Wireless Networks
(WiOpt ’05), pp. 294–304, April 2005.
[12] K. Zeng, W. Lou, and H. Zhai, “On end-to-end throughput
of opportunistic routing in multirate and multihop wireless
networks,” in Proceedings of the 27th IEEE Conference on
Computer Communications (INFOCOM ’08), pp. 1490–1498,
Phoenix, Ariz, USA, April 2008.
[13] MIT Roofnet, http://www.pdos.lcs.mit.edu/roofnet/.
[14] L. Qiu, Y. Zhang, F. Wang, M. K. Han, and R. Mahajan,
“A general model of wireless interference,” in Proceedings of
the 13th Annual ACM International Conference on Mobile
Computing and Networking (MobiCom ’07), pp. 171–182,
September 2007.
[15] K. Jain, J. Padhye, V. N. Padmanabhan, and L. Qiu, “Impact
of interference on multi-hop wireless network performance,”
in Proceedings of the 9th Annual International Conference on
Mobile Computing and Networking (MobiCom ’03), pp. 66–80,
September 2003.