Elsevier

Computer Communications

Volume 28, Issue 10, 16 June 2005, Pages 1138-1151
Computer Communications

Physical layer impact on the design and performance of routing and broadcasting protocols in ad hoc and sensor networks

https://doi.org/10.1016/j.comcom.2004.07.040Get rights and content

Abstract

Existing routing and broadcasting protocols for ad hoc networks assume an ideal physical layer model, where two nodes communicate if and only if they are at distance at most R, where R is the transmission radius. This article surveys our efforts to consider a more realistic physical layer model, and its impact on routing and broadcasting. We apply the log normal shadow fading model to represent a realistic physical layer to derive (accurately and approximately) the probability p(x) for receiving a packet successfully as a function of distance x between two nodes. We define the transmission radius R as the distance at which p(R)=0.5. We consider routing algorithms with and without acknowledgements, with messages consisting of one or more packets. For single packet routing with hop by hop acknowledgements, we propose a MAC layer protocol where receiver node acknowledges packet to sender node u times, where u×p(x)≈1. The expected hop count (EHC) between two nodes under this protocol is (1/p(x)2)+(1/p(x)) (for u=1). We show that forwarding to neighbor closest to destination is suboptimal, and that optimal forwarding distance is one that minimizes EHC per progress made, that is, minimizes h(x)=(1/p(x)2+1/p(x))(1/x). Depending on the power attenuation factor β (between 2 and 6), the optimal forwarding radius is between 0.7R and 0.8R. The function h(x) can be used to derive hop count optimal route discovery based routing. During route discovery, each node, receiving message from a neighbor at distance x (distance can be estimated from received signal strength), will set timeout proportional to h(x), before retransmitting. The destination will respond to the first message received. Neighbor discovery is not straightforward since hello messages are not received by all neighbors. We proposed several localized routing schemes for the case when position of destination is known, optimizing expected hop count (for hop by hop acknowledgement), or maximizing the probability of delivery (when no acknowledgements are sent). We then considered localized power aware routing schemes under realistic physical layer, when nodes can adjust their transmission powers. Finally, we discuss broadcasting in ad hoc network with realistic physical layer, and propose new concept of dominating sets to be used in broadcasting process.

Introduction

Wireless ad hoc networks [4], [5] emerged recently as a ‘hot’ research topic because of their potential applications in various situations such as battlefield, emergency relief, and conference environments. Sensor networks are currently one of prime research topics, and are being designed for monitoring environment for chemicals, temperature, movement, fire and other events. Ad hoc and sensor networks consist of hosts that communicate without a fixed infrastructure. Communications take place over a wireless channel, where each host has the ability to communicate with others in the neighborhood, determined by the transmission range. Since there is no infrastructure, every host has to determine its environment when the network is formed.

We assume that each node has a low-power Global Position System (GPS) receiver, which provides the position information of the node itself. If GPS is not available, the distance between neighboring nodes can be estimated on the basis of incoming signal strengths or time delays. Relative co-ordinates of neighboring nodes can be obtained by exchanging such information between neighbors [6].

In the routing task, a message is to be sent from a source node to the destination node. The nodes in the network may be static or mobile. The task of finding and maintaining routes in ad hoc networks is non-trivial since host mobility can result in unpredictable topology changes. Location updates schemes for efficient routing are reviewed in [7]. In a broadcasting task, a node wishes to send the same message to all the other nodes in the network.

We consider two routing scenarios. In one scenario, sender node is not aware of destination location, and initiates route discovery, which is a broadcasting task. In other scenario, source node is aware of geographic position of destination. Many routing algorithms proposed are non-local and require the complete knowledge and maintenance of the network topology. Recently, many localized routing algorithms have been proposed (a brief survey of them is given. in [8]), where nodes do not require the complete network topological information to perform the routing task. More precisely, nodes only require the position of themselves and their 1-hop neighbors (in some cases also position of their 2-hop neighbors), and position of destination. Consequently, neighboring nodes are aware of distances between them.

There are two assumptions about transmission power it is fixed and it is adjustable. In the first case, all nodes have a fixed and equal transmission radius R. Existing network layer protocols (with few exceptions) for ad hoc networks assume an ideal physical layer model, where two nodes communicate if and only if the distance between them is at most R. In this model, known as the unit graph model, two nodes within transmission radius can exchange correctly bits, packets and messages (we assume that messages are composed of few fixed length packets, and packets are composed of fixed length bit-strings). In the unit graph model there exists therefore the unique transmission radius at all layers of communication. We apply, however, log normal shadow fading model to represent a realistic physical layer. By applying a realistic physical layer, the notion of transmission radius needs to be carefully defined and properly used in algorithms. The packet reception probability p(x) depends on the probability b(x) of receiving a bit successfully and the length of the packet. There are different ways of determining R. Our approach is to divide message into fixed size packets, and transmit each packet individually. In this case, R can be determined so that packet error rate at distance R is 0.5. It obviously depends on packet length. The error rate for acknowledgements is then also 0.5 at distance R, since acknowledgements are assumed to be single packets with equal packet length, therefore the same probability for their reception is used. There are variety of ways to define medium access layer for acknowledging the packets. This interpretation for R appears to be the most convenient for deriving protocols and various acknowledgement schemes. When nodes can adjust their transmission radii, the goal is to minimize the total power for a route, with all retransmissions counted.

In this survey article, we consider routing with and without acknowledgements. In the HHR (Hop-by-hop retransmissions) model, a packet is retransmitted between two nodes until it is received and acknowledged correctly. We consider the separate HHR variant, where acknowledgements to the previous node and forwarding message to the next node are always done by separate messages. The variant where retransmissions to the next node can serve as acknowledgement to the previous node is left for future research.

For algorithms without acknowledgements, the probability of successful delivery of the packet (which is the product of all the probabilities of successful delivery of hops along the found route) from source to destination is used as a performance measure for routing algorithms.

Section snippets

Related work

There exists a vast amount of literature devoted to position based routing in ad hoc networks. Finn [9] proposed localized greedy scheme, where node, currently holding the message, will forward it to the neighbor that is closest to destination. Only nodes closer to destination than the current node are considered. Another milestone achievement is localized greedy-face-greedy (GFG) algorithm, proposed in [10], which guarantees delivery under ideal MAC layer and correct position information. It

The log-normal shadowing model

Most of the published results in ad hoc wireless routing and broadcasting are based on free-space or two-ray ground propagation models, which are simplistic and idealistic physical layer models. However, in real scenarios, the received signal strength is not only dependent on the distance between the transmitter and the receiver but also on the environment. Moreover, subsequent transmissions with the same transmission power, between same nodes in the same environment are not received with the

MAC layer protocol between two nodes

In this section, we consider HHR (hop-by-hop retransmission) routing protocol, where the sender of a packet requires the acknowledgement from receiver. To simplify our protocols and analysis, we assume that receiving node needs to send separate acknowledgement and forwarding packets to the previous and the next nodes on the route. We describe a simple MAC layer communication protocol between two nodes and present related analysis. After receiving any packet from sender, the receiver sends u

Optimal packet forwarding distance is less than transmission radius

In [17], we further show that the optimal packet forwarding distance to minimize the hop count is less than the transmission radius R. To derive this result, we place (n−1) equally spaced additional nodes, if needed and desired, between source S and destination D, along the straight line joining S and D. Let x=d/n be the distance between two consecutive nodes (see Fig. 4). We now derive the optimal values for n and x, by finding the expected hop count of such placement, and finding its minimum

Route discovery

We shall first consider the case when source node is not aware of the location of destination node, and not even aware of the location of its neighboring nodes. In coming sections, these assumptions will change. Qin and Kunz [11] already observed that, by applying reception signal thresholds, the delivery ratio and latency of selected route can decrease significantly. Position information was not used. In [18], we described several position based route discovery schemes. Based on distance from

Neighbor discovery

In the traditional unit graph model, the neighbor discovery is a trivial problem. Each node sends a hello message, which is received by all nodes at distance at most R. When physical layer model is considered, the problem becomes non-trivial, and more sophisticated protocols are needed.

Suppose that all nodes use equal power for transmission. We proposed s-hello protocol [19] as follows. Each node sends s hello messages to all its neighbors. The probability that a neighbor at distance x receives

Routing algorithms with acknowledgements

We now assume that each node is aware of its location, location of its neighbors, and position of destination node. Localized routing algorithms require only this information. In this section, we will consider routing with hop by hop acknowledgements (moreover, assuming that acknowledgement to previous node is made independently from forwarding message to the next neighbor on the route). We describe several localized routing protocols [17].

The Ideal Hop Count Routing (IHCR) [17] is based on the

Routing algorithms without acknowledgements

In the EER model, there are no hop-by-hop acknowledgements. When (and if) a message arrives at the destination, there may or may not be acknowledgments sent from the destination to the source node, as a routing task. We shall describe several localized routing protocols for EER model [20].

Consider the routing task of sending a message from source C to destination D. Consider intermediate nodes at distances x1, x2,…,xn. The probability that D receives the full message from C is p(x1) p(x2),…,p(xn

Power aware localized routing

The transmission power so far was assumed to be fixed. If nodes can adjust transmission power, a natural criterion is then to minimize the power needed for routing, instead of minimizing expected hop counts. Localized power aware routing schemes, for the unit graph model, were described in [15], and recently in [21]. Physical layer impact on power aware routing is considered in our work currently in progress [22], as follows.

So far we have used packet reception probability function p(x) for two

Routing multi-packet messages

In [23] we considered routing of messages that are composed of several packets. Assume that a message is divided into M packets, each packet sent separately.

We consider here only routing with acknowledgements in hop by hop fashion (HHR case). To simplify, we assume that with one acknowledgement packet, receiver may acknowledge several packets at once. Obviously this may not be possible if the message is very long (for instance, the number of packets is more than the packet length). We assume

Broadcasting

We shall now consider the task of broadcasting, assuming that each node is aware of the position of itself and its 1-hop (and possibly 2-hop) neighbors. The proposed solution [24] is based on the notion of physical layer based dominating sets, as follows.

Let A1,…,Ak be active neighbors of given node B, and let x1,…,xk be their distances to B. Then p(x1),…,p(xk) are packet probability rates. The probability q that at least one of packets from active nodes is received by B is then q=1−(1−p(x1))

Conclusion

To the best of our knowledge, this is the first study of position based routing and broadcasting in ad hoc network with a realistic physical layer. We investigated routing with and without acknowledgements, and presented several greedy routing algorithms for ad hoc wireless networks. We show that realistic physical layer does have impact on the choice of best localized schemes.

The localized nature of the protocols avoids the energy expenditure and communication overhead needed to build and

Ivan Stojmenovic received PhD degree in mathematics in 1985. He held regular or visiting positions in Serbia, Japan, USA, Canada, France and Mexico. He published over 170 different papers in journals and conferences, edited Handbook of Wireless Networks and Mobile Computing (Wiley, 2002), and co-edited Mobile Ad Hoc Networking (IEEE Press, 2004). His current research interests include wireless ad hoc, sensor and cellular networks. He is currently editor of several journals including Journal of

References (24)

  • F. Dai, J. Wu, An extended localized algorithm for connected dominating set formation in ad hoc wireless networks, IEEE...
  • I. Stojmenovic, M. Seddigh, J. Zunic, Dominating sets and neighbor elimination based broadcasting algorithms in...
  • J. Carle, D. Simplot-Ryl, Energy efficient area monitoring for sensor networks, IEEE Computer, 37, 2, February 2004,...
  • S. Basagni, M. Conti, S. Giordano, I. Stojmenovic (eds.), Mobile Ad Hoc Networking, IEEE Press/Wiley, 2004, to...
  • G. Giordano

    Mobile ad hoc networks

  • D. Niculescu, Positioning in ad hoc sensor networks, IEEE Networks, July 2004,...
  • I. Stojmenovic

    Location updates for efficient routing in ad hoc networks

  • I. Stojmenovic

    Position based routing in ad hoc networks

    IEEE Commmunications Magazine

    (2002)
  • G.G. Finn, Routing and addressing in large metropolitan-scale internetworks, ISI Research Report, ISU/RR-87-180, March...
  • P. Bose et al.

    Routing with guaranteed delivery in ad hoc wireless networks

    ACM DIAL M Workshop, Seattle

    (August 1999)
  • L. Quin et al.

    On-demand routing in MANETs: the impact of a realistic physical layer model

    Proceedings of the Second International Conference ADHOC-NOW

    (2003)
  • D. De Couto et al.

    A high-throughput path metric for multi-hop wireless routing

    Proceedings of the ACM Mobicom, San Diego

    (2003)
  • Cited by (50)

    • Randomized carrier-based sensor relocation in wireless sensor and robot networks

      2013, Ad Hoc Networks
      Citation Excerpt :

      Similar to the case in Fig. 6d we again notice a difference in the PR when comparing the cluster based variants against the non-cluster based variants. In the last decade, wireless sensor networks (WSNs) have attracted considerable research attention [3,15,28,30]. In this paper, we have investigated for the first time robot-assisted sensor relocation for coverage repair, that is, relocating deployed spare sensors to fix emerging sensing holes, and we have presented a localized solution, R3S2, its grid-based variant G-R3S2 and cluster-based variants C-R3S2 and C-G-R3S2.

    • Efficient broadcasting in multi-hop wireless networks with a realistic physical layer

      2013, Ad Hoc Networks
      Citation Excerpt :

      However, the algorithm cannot guarantee that nodes eventually receive the broadcasting message with probability no less than a given requirement, which is the focus of this work. Basically, the link reliability can be determined mainly by the Euclidean distance between the two nodes [23,32]. Besides, energy consumption in transmission from one node to another can also have a strong relationship with the distance between them according to the radio model given in [6,28,33].

    View all citing articles on Scopus

    Ivan Stojmenovic received PhD degree in mathematics in 1985. He held regular or visiting positions in Serbia, Japan, USA, Canada, France and Mexico. He published over 170 different papers in journals and conferences, edited Handbook of Wireless Networks and Mobile Computing (Wiley, 2002), and co-edited Mobile Ad Hoc Networking (IEEE Press, 2004). His current research interests include wireless ad hoc, sensor and cellular networks. He is currently editor of several journals including Journal of Multiple-Valued Logic and Soft Computing, IEEE Transactions on Parallel and Distributed Systems, Parallel Processing Letters, and Parallel Algorithms and Applications. He guest edited recently special issues in several journals including IEEE Computer Magazine (February 2004), IEEE Networks, and Wireless Communications and Mobile Computing.

    Amiya Nayak is an associate professor at the School of Information Technology and Engineering (SITE) at the University of Ottawa. He received his BMath degree in Computer Science from University of Waterloo and PhD in Systems and Computer Engineering from Carleton University, Ottawa in 1982 and 1991, respectively. He has over 15 years of industrial experience in software engineering in avionics and telecommunication applications, working at CMC Electronics and Nortel Networks prior to joining University of Ottawa in 2002. He has been an adjunct research professor at Carleton University since 1994 and the Canadian Editor and the Book Review Editor of VLSI Design: An International Journal of Custom-Chip Design, Simulation, and Testing since 1998. His research interests are in the areas of fault-tolerant computing, ad hoc networks, and distributed computing.

    Johnson Kuruvila has 15 years of industry experience in Research, S/W development, Embedded/Distributed System Design and IP Networking/Wireless protocols. Earlier he was Lead Engineer at AcceLight Networks and Director of Routing/Signaling Platforms at AcceLight Research. He was Manager of IP Routing system development at the Optera Packet Core project in Nortel Networks. He has been involved with the development of many leading edge products and is a recipient of many performance awards and has given lectures to many product groups in IP Routing Protocols, MPLS/GMPLS signalling and Parallel/Distributed Algorithms. He has given more than 10 invited lectures at various research labs and supervised many graduate and undergraduate research works at various universities. He holds degrees in Electronics/Communication Engineering and Computer Science. He has published many papers and has filed two patents. Currently, he is doing his research in Wireless ad hoc networks at University of Ottawa.

    Francisco Ovalle-Martinez was born in Mexico City in 1977. He earned his Bachelor of Science degree in Telecommunications Engineering from the Autonomous National University of Mexico (UNAM), in 2001. Francisco has had the opportunity to develop and enhance, his professional skills through various software and hardware firms in Mexico, such as Red de Control Corporativo and Yum International Restaurants Mexico. His tenacity, determination and ever present desire for success, have taken him through different paths; today, he is on his way to achieve a master degree in Computer Science at the University of Ottawa, Canada. He has been awarded and holds a CONACyT grant. His interest areas mainly include mobile computing, digital communications, data networks and security.

    Eduardo Villanueva-Pena graduated (with honours) from the National Autonomous University of Mexico in 2001, with the BCS degree. In 2002, he received the MSc degree in Engineering from the University of Sheffield, UK. He was an associated researcher at the Complex Systems Laboratory (Edo. Mex., Mexico) from 1998 to 2000, at the University of the Mexican Valley. He won a third prize at Mexico City Informatics Olympiad in 1998. He got the third best grade point average at the end of his bachelor studies. Currently, he is doing his PhD in Engineering as a full time student at the University of Ottawa. Mainly, his research interest resides on wireless networks.

    View full text