3.5 Case Studies
3.5.2 A Cluster-Based Network with Sleep Mode Operation
Last but not the least, we consider a cluster-based network. Cluster-based routing protocols have been studied in great detail; e.g., [79–82]. Generally, in a cluster-based network, nodes in a local area elect a “community center”. Leaf nodes forward data to the center nodes. A center node summarizes its local data and forms a condensed packet to forward on. Conversely, a center node is also responsible for distributing remote messages to its leaf nodes.
Here we consider a cluster-based network whose structure was determined using the Energy-aware Virtual Backbone Tree (EVBT) [82] algorithm. The EVBT algo-rithm is proposed under the following node energy consumption model. Given a radio transmission range d, the energy spent to transmit a message, Etx, and the energy spent to positively receive a message, Erx, are the following, respectively:
Etx = α11+ α2d2, Erx = α12.
In [82], nodes are classified into tree nodes and leaf nodes. A leaf node associates itself with a tree node closest to it. A message originated from a leaf node is first
Table 3.2: The node energy profile for EVBT simulation.
Cause Symbol Expression
Energy to transmit a packet Etx α11+ α2d2 Energy for negative detection Emx α3
Energy for positive detection Erx α12
Table 3.3: The node energy profile for the EVBT-approximate RWN simulation.
Cause Symbol Expression
Energy to transmit a packet Et α2d2 Energy for negative detection Er α3
Energy for positive detection Es 0
transmitted to the corresponding tree node. Then the message is progressively relayed among tree nodes until it reaches the root, which is the sink node.
The originally proposed EVBT network does not include a sleep mode operation, so we add the following sleep mechanism to it. In the network, the wakeup period is Td. Tree nodes wake up periodically to detect incoming traffic either from a leaf node or from a neighboring tree node. If a leaf node has an original message to send, it wakes up during a sleep epoch to send the message to its tree node. Otherwise, a leaf node never wakes up. A message can advance one hop per wakeup period. The goal for a network designer is to pick the optimal values for the radio transmission range d, the number of tree nodes, and the wakeup period Td.
For this network, an analogous RWN has the following parameters:
• transmission range d
• duty-cycle period Td
• the wakeup density equals the geographical density of tree nodes
Under this transformation, the original optimization problem corresponds to the case where p, D, and Td are all optimize-able in the analogous RWN.
We simulate a network which, except for the sleep behavior, is identical to the one in [82]. Specifically, we simulate a network with 10,000 nodes uniformly placed in a 600 meter-by-600 meter area. Every node originates a message to transport to
mean delivery speed (m/s)
optimalnumberoftreenodes
α3=11.52 µJ/packet in EVBT α3=11.52 µJ/packet in RWN α3=1.28 µJ/packet in EVBT α3=1.28 µJ/packet in RWN
101 102 103
101 102 103
Figure 3-13: The optimal number of tree nodes as a function of mean delivery speed in the EVBT network with sleep mechanism adopted. α11 = 12.8 µJ/packet. α12 = 12.8 µJ/packet. α2= 16nJ/packet/ m2.
the sink node at the center of the network. The average message origination period is 1 second. The node parameters for both the EVBT network and the analogous RWN are shown in Table 3.2 and Table 3.3 , respectively. Please note that, in order to render this optimization problem meaningful, a small positive energy is consumed when a tree node detects no messages.
The optimal number of tree nodes that minimizes network energy consumption as a function of mean delivery speed is shown in Figure 3-13. For comparison purposes, the counterpart for the analogous RWN, assuming that α11 = 0 and α12 = 0, is shown as well. From the figure, we can observe that, despite the mismatch in node power consumption model, the RWN predicts well the optimal settings for this cluster-based WRN.
3.6 Summary
In this chapter, we have investigated the trade-offs between network energy consump-tion and message forwarding delay in random wakeup wireless networks. We have proposed the random wakeup network with opportunistic relay as a framework in our
analysis. We first derived the analytical model for network power consumption in the random wakeup network, and then we apply the model in an optimization problem to find the optimal design parameters (transmission range, duty-cycle period, and wakeup density) that can minimize the network power consumption while meeting the message delay requirement. Finally, we presented a case study to show that the proposed framework and analysis reasonably models the power consumption behavior of wireless relay networks with random sleep-awake schedule.
Several interesting properties that we have found in this chapter include: (1) The optimal number of nodes participating in the relay activities in a duty-cycle period within the transmission range is not a function of the allowable delivery delay nor of how often a message is originated. Instead, it depends only on the path loss exponent and the response-to-transmit energy ratio. (2) When the optimal routing and sleep parameters are applied, the proportion of power consumption for detecting potential incoming messages in the whole network is α−1α+2, indicating that the power consumed for detection should be of the same order as the power spent in transmissions. (3) The minimal network power consumption grows at the α−1α+2-th power of the factor by which the minimal required delivery speed increases. That is, to reduce the message delivery delay in RWNs, the investment of network energy can be quite efficient.
THIS PAGE INTENTIONALLY LEFT BLANK
Chapter 4
Sleep Mode Operations for Information Exchange and
Information Dissemination in
Green Wireless Multi-Hop Relay Networks
A wireless relay network (WRN) [83] is an aggregation of nodes with the ability to transmit and receive wirelessly. To serve the underlying application, nodes may wish to communicate with each other. Whenever necessary, nodes serve as relays to transport messages for others.
For a WRN with battery-powered nodes, the ideal use of power is clearly of great significance. In many sensing applications, the node energy used for sensing is mi-nuscule compared to that used for wireless communication [48]. As a result, the way nodes communicate with each other shall be carefully designed.
Communication between nodes requires three actions: transmission, detection, and reception. Traditionally, network power consumption is optimized to minimize the total transmission energy. However, the reception mode power which we define
as the sum of energy to detect messages and receive such messages may easily rival the transmission energy [84], especially in low traffic density WRNs.
To reduce the energy consumption in reception mode, a popular approach is to switch the node into sleep mode in which a node shuts off its radio transceiver to disengage itself from the network for a period of time [80]. Compared with the power consumption in transmission mode or reception mode, the power consumption in sleep mode is often negligible.
In general, communication paradigms for energy-efficient WRNs can be charac-terized into: one-to-one [85], to-one [86, 87], one-to-many [88, 89], and many-to-many [90]. The optimization of power consumption often highly depends on these paradigms which the underlying network applications belong to. In these paradigms, information exchange in which many nodes wish to exchange its own message with other nodes in the network is essential for various WRN applications, e.g., consensus achievement [91]. Such a communication pattern often depletes a large amount of network energy, however, it is rarely investigated.
In this chapter, we investigate the optimization of power consumption minimiza-tion for informaminimiza-tion exchange which is a dual problem for message delivery in Chapter 3. We desire to seek for general guidelines in the design of energy-efficient WRNs. In the optimization problems, the power profile of nodes, the path loss exponent, and the message origination period are given and the position of nodes are determined to meet the sensing needs. Three key parameters are under our control: the transmis-sion range, the duty-cycle period, and the participation density. The optimization requirement is that all interested messages shall be exchanged in a targeted region within the allowable delay time.
We propose a “random gossip” wireless-relay network (RGN) as a framework for our analysis. In the RGN, we consider that each node initiates a message and wishes to exchange its message with all the other nodes in the network. We will demonstrate that our results in the RGN can indeed predict the optimal value of operational parameters for information exchange as well as information dissemination in actual WRNs.
We find a number of properties for the three design parameters in the optimized RGNs. First, a key relationship is that the optimal number of nodes particiating to broadcast messages within the transmission range in a duty-cycle period depends only on the path loss exponent and the network dimensionality. It does not depends on factors such as the physical network size, the detection energy, and the message origination rate. Next, the existing optimal number of broadcasting nodes in a time epoch within the transmission range shows that either shortest hop transmission or blind flooding is rarely optimal for power consumption minimization. Last but not the least, it is shown that neither increasing nor decreasing the physical network size will affect the optimal value of these design parameters. The optimal setting for power consumption minimization is a scalable solution.
The rest of this chapter is organized as follows. In Section 4.1, we present the random gossip framework with periodic listening. In Section 4.2, we analyze the power consumption in the random gossip WRN. In Section 4.3, we derive optimal value of the design parameters for minimizing overall network power consumption.
In section 4.4, we apply our results for information exchange as well as information dissemination in proposed WRNs. Summary of the chapter is given in Section 4.5.