Effective and efficient neighbor detection for proximity-based mobile applications
Introduction
With the increasing use of mobile devices and particularly smartphones, we face the emergence of a new blend of distributed applications known as Proximity-Based Mobile (PBM) applications [10], [11], [12]. These applications enable a user to interact with others in a defined range and for a given time duration e.g., for social networking (WhosHere [53], LoKast [31]), gaming (Bluetooth gaming apps [8]) and driving (Waze [52]).
Discovering who is nearby is a basic requirement of various PBM applications. In a simple usage scenario of social networking applications such as WhosHere [53] or LoKast [31], a user can discover other users in a defined range, view their profiles and chat with a user or a group of users with her phone. Usually, the highly dynamic nature of these applications (which is basically due to the mobility of devices) makes neighbor detection time-constrained, i.e., even if a device remains in proximity for a limited amount of time, it should be detected with a high probability as a neighbor. In addition, the limited battery life of mobile devices requires the neighbor-detection to be performed by consuming as little energy as possible.
In this paper, we consider the following problem: how can a device be detected by another device with both maximum effectiveness and maximum efficiency, given a maximum amount of time that they remain in proximity of each other? If not, how can an effectiveness-efficiency tradeoff be made? Effectiveness refers to the degree to which the detection is successful and is measured by the detection probability, while efficiency refers to the degree to which the detection is energy saving and is measured by the inverse of energy consumption per device. To address this problem, we evaluate effectiveness and efficiency in a single-hop mobile ad hoc network (MANET). The evaluations are performed under realistic assumptions and based partly on simulations using the ns-2 [37] network simulator.1
There are two main reasons behind our choice of a MANET as the underlying network architecture. Firstly, MANETs seem to be the most natural existing technology to enable PBM applications. In fact, similarly to PBM applications, in a MANET two nodes can communicate if they are within a certain distance of each other (to have radio connectivity) for a certain amount of time. Secondly, mobile devices are increasingly equipped with ad hoc communications capabilities (e.g., WiFi in ad hoc mode or Bluetooth) which increases the chance of MANETs to be one of the future mainstream technologies for PBM applications.
Since the quality of radio signals (and consequently the detection probability) is affected by the environment attenuation, for our study we consider three typical urban environments where PBM applications are used, i.e., indoor with hard partitions (corresponding to offices with thick walls), indoor with soft partitions (corresponding to exhibitions with temporary partitions) and outdoor urban areas (corresponding to a music festival in downtown). To simulate these environments, we use a radio propagation model known for modeling the obstructed urban environments called Log-Normal Shadowing (LNS).
In our study, a node periodically broadcasts a hello message during a fixed time interval in order to be detected as a neighbor. We assume that the nodes use the IEEE 802.11a standard for the physical and MAC layer. Thus, we study the impact of two key parameters that influence effectiveness and efficiency, i.e., the transmission power and the time interval between two consecutive broadcasts. In performing the evaluations, we are particularly interested to answer the following questions:
- •
In each environment, when does a change in the value of any of the above mentioned parameters increase effectiveness and efficiency, or on the contrary, when does it deteriorate them?
- •
In each environment, is there a unique combination of these parameters that could maximize both effectiveness and efficiency? If not, how could a tradeoff between effectiveness and efficiency be made?
This paper is, to the best of our knowledge, the first study on the impact of transmission power and broadcast interval on effectiveness and efficiency of neighbor detection for MANETs in urban environments. It provides a detailed simulation study and defines the metrics that can be used to interpret the results. In order for our results to be close to reality, the study is performed under realistic assumptions. For one thing, we use 802.11a technology for communication between nodes and we assume a probabilistic radio propagation model for urban environments. Furthermore, we calculate the energy consumption using the specification of typical smartphones.
The remainder of the paper is as follows. In Section 2, we describe our system model. In particular, we define the neighbor detection algorithm, which takes transmission power and broadcast interval (this pair constitutes a strategy) as input. In Section 3, we formulate the problem studied in this paper. It basically consists of finding the most effective and the most efficient strategy in each environment. If these strategies are not equal in an environment, we intend to find a strategy that makes a reasonable tradeoff between effectiveness and efficiency. We also define the set of strategies for which the effectiveness and efficiency are evaluated. In Section 4, we evaluate the effectiveness for the set of predefined strategies. We also discuss the impact of changing transmission power and broadcast interval on effectiveness. Finally, we identify the most effective strategy in each environment. In Section 5, we evaluate the efficiency for the set of predefined strategies. We show that efficiency is independent of the environment and we discuss the impact of changing transmission power and broadcast interval on efficiency. Finally, we identify the most efficient strategy. In Section 6, we compare the results of Sections 4 Evaluation of effectiveness, 5 Evaluation of efficiency. We observe that we cannot find a strategy that maximizes both effectiveness and efficiency in any environment. The reason is that, regardless of environment, effectiveness and efficiency are in conflict with each other. We then propose an approach to make a tradeoff between effectiveness and efficiency. Using this approach, we find the tradeoff strategy in each environment and we show that it has a relatively good effectiveness and efficiency compared to other strategies. Finally, we discuss related work in Section 7 before concluding in Section 8 with a perspective on future work.
Section snippets
System model
In this section, we present the system model, and whenever necessary, we describe the reasons behind our modeling choices.
Problem statement
We characterize neighbor detection by two main aspects:
- •
Effectiveness is defined as the degree to which the neighbor detection is successful and is measured by the detection probability. Thereby, maximizing effectiveness boils down to maximizing the detection probability.
- •
Efficiency is defined as the degree to which the detection is energy saving and is measured by the inverse of energy consumption per process. Thereby, maximizing efficiency boils down to minimizing the energy consumption per
Evaluation of effectiveness
Effectiveness is measured by the neighbor detection probability. Thus, in this section we first describe our approach to calculate the detection probability of each strategy, which is based on simulations. Then, we present our simulation setup and the results. In particular, while presenting the results, we discuss how a change in or can affect the detection probability in each environment. We also define two packet dropping metrics that we use to interpret the results. Finally,
Evaluation of efficiency
Efficiency is measured by the inverse of energy consumption per process. Therefore, in this section we first define a model of energy consumption and then design an algorithm that, based on the model, calculates for each strategy the energy consumption per process. After describing the algorithm, we present the results and we discuss how a change in or can affect the energy consumption. Finally we compare the efficiency of the strategies and present the most efficient strategy.
Effectiveness-efficiency tradeoff
In this section, we first compare the results of effectiveness and efficiency evaluations to find the strategy that maximizes both effectiveness and efficiency for each environment. We show that there is a conflict between effectiveness and efficiency. Hence, such a strategy does not exist in any environment. We then propose an approach to make a tradeoff between effectiveness and efficiency and we find the tradeoff strategy for each environment. Finally, to show how good the tradeoff strategy
Related work
The existing studies on transmit power control in ad hoc networks [27], for the most part, consider the unicast communications and therefore are not relevant to our work. Also, most of the papers which study the problem of reliable broadcast in MANETs, consider one-shot broadcast and not periodic broadcast [21]. For these reasons, in this section we only discuss the works performed in the two fields which we believe are the closests to our work, i.e., enhancements of the hello protocol; and
Conclusion
To the best of our knowledge, this is the first paper that studies the impact of transmission power and broadcast interval on effectiveness and efficiency of neighbor detection for MANETs in different urban environments. Our results can be used as a basis to design adaptive neighbor detection algorithms for urban environments. Such algorithms can adapt the transmission power and broadcast interval based on environment and application guarantees on effectiveness and efficiency. To deploy such
Acknowledgment
This research is partially funded by the Swiss National Science Foundation in the context of Project 200021-140762.
Behnaz Bostanipour is a PhD student in computer science at the University of Lausanne in the Distributed Object Programming (DOP) Lab under the supervision of Prof. Garbinato. Before joining the DOP Lab, she obtained a BSc. degree and a MSc. degree in Communication Systems from Swiss Federal Institute of Technology of Lausanne (EPFL). Her research interests focus on mobile ad hoc networks and distributed computing. In particular, she studies and designs new abstractions and algorithms for
References (56)
- et al.
Context-aware broadcasting approaches in mobile ad hoc networks
In Comput. Netw.
(2010) - IEEE 802.11: Wireless LAN MAC and Physical Layer Specifications,...
- IEEE std 802.11a/D7.0-1999, part11: wireless LAN medium access control (MAC) and physical layer (PHY) specifications:...
- Official IEEE 802.11 working group project timelines, 2014....
- Apple iPhone 4S Teardown....
- et al.
A Survey of Mobility Modeling and Analysis in Wireless Ad Hoc Networks
(2006) - M. Bakht, M. Trower, R. Kravets, Searchlight: helping mobile devices find their neighbors, in: ACM SIGOPS Oper. Syst.,...
- et al.(2013)
- Best iPhone bluetooth games....
Cost-Benefit Analysis: Concepts and Practice
(2006)
Quorum-based asynchronous power-saving protocols for IEEE 802.11 ad hoc networks
ACM Mob. Netw. Appl.
Principles and protocols for power control in wireless ad hoc networks
IEEE JSAC
Cited by (3)
A neighbor detection algorithm based on multiple virtual mobile nodes for mobile ad hoc networks
2017, Computer NetworksCitation Excerpt :In ad hoc networks neighbor detection is usually studied as a building block for applications such as routing, leader election, group management and localization. The majority of the existing neighbor detection algorithms belong to the hello protocols family [1,4,6,17,24,26,28,37,50]. They are based on the basic hello protocol first described in Open Shortest Path First (OSPF) routing protocol [38], which works as follows: each node in the network periodically sends hello messages to announce its presence to close nodes, and maintains a neighbor set.
Neighbor Detection Based on Multiple Virtual Mobile Nodes
2016, Proceedings - 24th Euromicro International Conference on Parallel, Distributed, and Network-Based Processing, PDP 2016
Behnaz Bostanipour is a PhD student in computer science at the University of Lausanne in the Distributed Object Programming (DOP) Lab under the supervision of Prof. Garbinato. Before joining the DOP Lab, she obtained a BSc. degree and a MSc. degree in Communication Systems from Swiss Federal Institute of Technology of Lausanne (EPFL). Her research interests focus on mobile ad hoc networks and distributed computing. In particular, she studies and designs new abstractions and algorithms for proximity-based mobile applications running on mobile devices and smartphones.
Benoît Garbinato is a professor in computer science at the University of Lausanne since 2004, where he leads the Distributed Object Programming (DOP) Lab. In the nineties, he contributed to the emerging research trend on separation concerns and protocol composition in fault-tolerant distributed systems, as part of his Ph.D. thesis. He then worked in the industry, first for the research lab of UBS in Zurich (Ubi-lab), where he lead the software engineering group, and later for Sun Microsystems professional services, as senior software architect. Since his return to the academic world, his research and teaching activities focus on the design and implementation of adequate programming abstractions for emerging distributed architectures, such as pervasive and mobile systems. He has over 50 publications in international conferences and journals, and is member of the ACM and the IEEE societies.