Multiple-metric hybrid anycast protocol for ... - Semantic Scholar

Download PDF
0 downloads 0 Views 4MB Size Report
Comment
discovery. Teresa A. Dahlberg is a Professor of Computer Science at the University of North Carolina at Charlotte. Her research on wireless networks addresses ...
Int. J. Ad Hoc and Ubiquitous Computing, Vol.

Multiple-metric hybrid anycast protocol for heterogeneous access networks Lijuan Cao* Department of Computer Science and Engineering, Johnson C. Smith University, Charlotte, NC 28216, USA E-mail: [email protected] *Corresponding author

Kashif Sharif, Teresa A. Dahlberg and Yu Wang Department of Computer Science, University of North Carolina at Charlotte, Charlotte, NC 28223, USA E-mail: [email protected] E-mail: [email protected] E-mail: [email protected] Abstract: The wireless multihop to an Access Point (AP) model appears to be a promising component of future access network architectures. A key challenge is managing diverse resources at APs while discovering efficient multi-hop paths from source to AP based on selection criteria dictated by applications or necessitated by network constraints. We propose a new hybrid proactive/reactive anycast routing protocol that integrates multiple-metrics to calculate path cost and selects the appropriate AP. Simulation analysis shows that our approach outperforms single-metric protocols while supporting flexible service criteria, including load balancing at APs. Keywords: anycast routing; multiple metrics; heterogeneous networks; hybrid routing. Reference to this paper should be made as follows: Cao, L., Sharif, K., Dahlberg, T.A. and Wang, Y. (2011) ‘Multiple-metric hybrid anycast protocol for heterogeneous access networks’, Int. J. Ad Hoc and Ubiquitous Computing, Vol. Biographical notes: Lijuan Cao is an Assistant Professor in the Department of Computer Science and Engineering at Johnson C. Smith University. Her research interest includes wireless sensor networks, peer-to-peer networks, heterogeneous access networks, resource management and routing algorithm design. She received PhD from University of North Carolina, Charlotte in 2008, and BS Degree from University of Electronic Science and Technology in 2003. Kashif Sharif is a PhD student in Department of Computer Science at the University of North Caroline at Charlotte. His research focus is on routing and sink-discovery protocols in application specific mobile ad hoc and sensor networks, path optimisations and resource discovery. Teresa A. Dahlberg is a Professor of Computer Science at the University of North Carolina at Charlotte. Her research on wireless networks addresses resource management protocols, data management for sensor networks, and analytic, simulation and experimental modeling and analysis techniques. She received the MS and PhD in Computer Engineering from North Carolina State University and the BS in Electrical Engineering from the University of Pittsburgh. Yu Wang received the PhD Degree in Computer Science from Illinois Institute of Technology in 2004, the BEng Degree and the MEng Degree in Computer Science from Tsinghua University, China, in 1998 and 2000, respectively. He is an Associate Professor of Computer Science at the University of North Carolina at Charlotte. His research interests include wireless networks, ad hoc and sensor networks, mobile computing, and algorithm design. He is a member of the ACM and a senior member of the IEEE, and IEEE Communications Society.

Copyright © 2011 Inderscience Enterprises Ltd.

37

Int. J. Ad Hoc and Ubiquitous Computing, Vol. 8, Nos. 1/2, 2011

1 Introduction The vision of future generation networks is evolving towards one that includes interoperable heterogeneous wireless access technologies to provide seamless access to core networks. Today’s markets include a proliferation of cellular, WiFi and WiMax technologies for access to telecom, internet and entertainment networks via mobile devices such as phones, PDAs, laptops and sensors. Mobile devices equipped with multiple interfaces are already available for consumer use. Recently, there have been active efforts to combine the advantages of cellular and ad hoc wireless access modes, and exploit their added benefits for system performance (Bhargava et al., 2004; Hsieh and Sivakumar, 2002; Liu et al., 2003; Ye et al., 2003; Wu et al., 2001; Zhou and Yang, 2002; Aggelou and Tafazolli, 2001; Wu et al., 2000; Lakkavalli et al., 2003; Yanikomeroglu, 2002; Kubisch et al., 2003). Much of the proposed solutions comprise incremental changes to cellular resource management protocols or to ad hoc routing protocols as a way to extend cellular to ad hoc or vice versa. Here we take a more comprehensive approach by integrating route discovery, access point discovery, and access point load balancing into a flexible multiple-metric hybrid routing protocol for heterogeneous wireless access networks. Our design is based on our conclusions that future access networks will include heterogeneous air-interfaces, must support diverse applications, and will have varying network resources and constraints. We therefore employ an anycasting routing paradigm, combined proactive and reactive route discovery, and multiple path cost metrics to support flexible cost-performance decisions during route and access point discovery. To explain, first consider that within future generation networks, mobile users may be able to access multiple Access Points (APs) for connection to the internet. Thus, it is important for mobile users to locate the best AP from “one or more of a group” of APs, which can be better modelled by an anycast or manycast communications paradigm, rather than unicast or multicast. Here, the notion of ‘best’ can be described as optimum for communication based on some selection criteria. As different applications might have different requirements, moreover, the cost of providing services to users varies from one AP to another, as determined by a complex combination of issues including available bandwidth, channel capacity, service availability, etc. The decision of AP selection should be determined by routing method based on both application requirement and provided services from APs. Though anycasting is originally an internet service for best effort delivery of datagrams from a host to at least one, and preferably only one server from the nearest ‘group of servers’, it has been applied to routing protocol design for wireless ad hoc and sensor networks (Thepvilojanapong et al., 2005; Hou et al., 2005; Lenders et al., 2006; Intanagonwiwat and Lucia, 1999; Wang

Copyright © 2011 Inderscience Enterprises Ltd.

et al., 2003; Peng et al., 2005). Wang et al. (2003) proposed anycast routing protocol based on Ad Hoc On Demand Distance Vector (AODV) (Perkins et al., 2003). Anycast routing is supported by introducing a 4-bit Anycast Group ID, which is contained in the Route Request (RREQ) message along with other flags for discovery of the nearest anycast service provider. This protocol is designed to work purely in ad hoc environments for evenly distributing the load on different available anycast server nodes. In Peng et al. (2005), to support anycast service, Dynamic Source Routing (Johnson and Maltz, 1996) Protocol is extended with a similar idea as Wang et al. (2003) for anycast ID or Anycast address. Most existing routing protocols for multihop wireless networks are simply designed using reactive schemes, meaning that route discovery is initiated on an as-needed basis, for an initial connection, or when an existing route breaks. In contrast, proactive routing protocols discover and maintain routes before they are needed. Proactive route discovery minimises the time needed for path selection, but requires additional overhead to maintain routes that are not being used. Because of the need to minimise energy consumption and because of mobility, most ad hoc routing protocols are reactive. However, a unique feature of routing in heterogeneous wireless networks, as compared to ad hoc routing, is that only APs serve as possible ‘destinations’ to the multi-hop path within the access network. These destinations are fixed and have specific access functionality. This calls for a fresh look at the tradeoffs in using proactive or reactive route discovery policies. Besides hybrid anycast routing paradigm, the calculation of path cost is also a critical component of routing design for heterogeneous access networks. In wireless networks, devices are usually resource constrained, e.g., limited battery capacity, buffer space, CPU processing capability, memory size, etc. Thus, the path cost metric, which guides path selection and resource consumption, is a crucial element of the protocol design. Prior work on multi-hop wireless routing protocols relies largely on the use of single cost metric, such as number of hops in a path (Perkins and Royer, 1999; Perkins and Bhagwat, 1994), the energy consumed along a path (Singh and Raghavendra, 1998; Singh et al., 1998; Kim et al., 2002), the energy remaining after using a path (Toh, 2001), the load carried by nodes along a path (Lee and Gela, 2001; Wu and Harms, 2001; Hassanein and Zhou, 2001; Saigal et al., 2004), etc. The aim of such protocols is to guide path selection to favour the least-cost path, where the path cost metric reflects the criteria that is to be minimised. However, this approach does not suffice for future access networks for at least two reasons: first, applications might have multiple quality of service requirements that must be simultaneously considered during route discovery; and second, development of new radio access and wireless communication technologies is producing a wide array of wireless devices, having different levels of constrained

38

L. Cao et al.

resources. E.g., a laptop might have more powerful process capability as compared to a cell phone. In recent research, there has been some work focusing on optimal AP selection in IEEE 802.11 WLANs (IEEE Standard, 2008). In Vasudevan et al. (2005), Vasudevan et al. defined potential bandwidth as a metric to determine AP selection, which can be calculated based on delays experienced by beacon frames from an AP. In order to achieve load balancing among multiple available APs, Chen et al. (2006) presented two algorithms to estimate the traffic load at APs by observing the IEEE 802.11 frame delays and used the result to determine the AP selection. In order to more efficiently utilise the WLAN resource, Abusubaih et al. (2006) derived a metric based on traffic class generated by the mobile node. In addition it takes into account the traffic being generated by nodes in the same cell. The authors have shown through simulation that utilising multiple resource parameters, better throughput can be achieved. Rather than using certain specified metric to make the AP selection decision, there is also some work using optimisation methodologies, for example, (Akl and Park, 2005; Lee et al., 2009) formulated constrained optimisation problems to achieve optimal AP selection and traffic allocation in the network. While these work focus on a single hop environment, the prime difference of our work is the multihop nature. Moreover, rather than using a single metric to determine the AP selection, our architecture has the ability to aggregate and transport multiple metrics for both path and AP selection. Nonetheless, these metrics and algorithms can be adopted to work in our architecture easily. In this paper, we describe our hybrid anycast routing protocol that integrates various cost metrics to guide path selection. Our hybrid mechanism divides the multihop portion of the access network into a proactive and a reactive region. The proactive region enables APs to advertise their services by maintaining state information at mobiles or relays within close proximity of the APs. The reactive region enables mobiles to discover APs, as needed, by interrogating nodes in the proactive region. A combination of proactive and reactive routing reduces communication overhead and delay, while increasing throughput. The use of multiple cost metrics and anycast routing paradigm provides flexible support of service and resource requirements. As our proposed solution is a general framework of hybrid anycast protocol, it can be applied to a wide range of heterogeneous access networks (any wireless networks with multihop relay to access points). These networks may include combinations of cellular, Wimax, WiFi, mesh networks, sensor networks and other emerging personal and long distance communication standards, such as those systems in Bhargava et al. (2004), Hsieh and Sivakumar (2002), Liu et al. (2003), Ye et al. (2003), Wu et al. (2001), Zhou and Yang (2002), Aggelou and Tafazolli (2001), Wu et al. (2000), Lakkavalli et al. (2003), Yanikomeroglu (2002) and Kubisch et al. (2003).

This paper is organised as follows. Section 2 describes the system model and assumptions that we use for analysis. Section 3 provides the details of our proposed multiple-metric hybrid protocol while Section 4 presents some issues in the implementation. Simulation results are presented in Section 5. Section 6 concludes the paper. A preliminary conference version of this paper appeared in Cao et al. (2009). This version contains new analysis of optimal proactive radius, new experimental results, and better overall presentation.

2 System model In this paper, we assume a general heterogeneous network architecture as shown in Figure 1. There are two basic entities in the system: Mobile Nodes (MNs) and Access Points (APs). MNs are mobile devices which may have multiple interfaces (e.g., 3G, 802.11, or 802.16) as well as the capability to relay traffic between interfaces. APs are physical access points that connect MNs to the core network and terminate the wireless portion of the network. Different APs can use various technologies. We assume existing protocols or system designs (Bhargava et al., 2004; Hsieh and Sivakumar, 2002; Liu et al., 2003; Ye et al., 2003; Wu et al., 2001; Zhou and Yang, 2002; Aggelou and Tafazolli, 2001; Wu et al., 2000; Lakkavalli et al., 2003; Yanikomeroglu, 2002; Kubisch et al., 2003) are available to integrate heterogeneous access technologies into the core network. We will only focus on AP discovery and path selection in the multihop part of the architecture. We also assume that all nodes have prior information about anycast group memberships and identities, using protocols available in the literature. Figure 1 Architecture of heterogeneous access network (see online version for colours)

3 Multiple-metric hybrid anycasting protocol We derive our multiple-metric hybrid anycasting protocol from the AODV architecture; with considerable

39

Multiple-metric hybrid anycast protocol modifications to support anycasting, distributed regions, and multiple cost metrics. An objective of our protocol is for a mobile node to establish connection with an AP in an anycast group based on multiple path cost metrics, Thus, the selected AP can forward packets to the destination in the core network.

•

Route discovery (Reactive region). If an MN does not have any valid route available to any member of the anycast group in its routing table, it broadcasts a RREQ. Upon receiving the RREQ, an intermediate node first checks whether it has received this RREQ before. If yes, it drops the RREQ. Otherwise, it updates the hop_count entry by adding one, and updates the cost field with equation (8). The intermediate node then creates a new entry in its routing table to record the previous hop and rebroadcasts the RREQ. RREP can only be generated by AP members of the anycast group or MNs in proactive regions that have an active path to any member of the anycast group. Therefore, upon they receive the first RREQ message, they check whether the AP’s capacity can satisfy the requirement in the RREQ message. If the requirement is satisfied, RREP is generated and sent back to the source along the reverse path; otherwise, RREQ message will be dropped. For later RREQ messages, RREP is only generated for those with smaller path cost value. Upon receiving the first RREP, an intermediate node records the previous hop and relays the packet to the next hop. Similarly, later RREP message will be forwarded only if it has a smaller path cost value.

•

Route selection. Route selection is related to the cost metric used in the protocol, i.e., AODV selects the path with the first RREP. While using multiple-metric included in the RREQ, our anycast protocol selects the route with the smallest cost value out of all received RREPs. After the source node receives the first RREP, it starts sending out data. It will switch to the other path only if the cost value in the corresponding RREP is smaller.

•

Route maintenance. Route maintenance is the same as for classical AODV.

3.1 Network regions Hybrid routing has been studied in wireless ad hoc networks. It leverages the tradeoff between the reduced delay provided by a proactive approach with the reduced communications overhead provided by a reactive approach. In order to support both proactive and reactive approaches, the network is divided into two regions: •

Proactive region. APs and MNs within an m hop radius of an AP are in the proactive region. All MNs maintain active information about AP in this region through periodic Hello packets sent by AP. Hereafter, we call m the proactive radius.

•

Reactive region. All MNs that more than m hops away from an AP are part of the reactive region, and use a reactive anycast routing protocol to discover routes to an AP.

The objective of our hybrid anycast protocol is for a mobile node to establish communication with an AP in an anycast group so that the selected AP can forward packets to the destination in the core network. The route to destination for all data packets is selected through any of the AP based on a decision metric. APs are entry points for MNs to access the internet and are part of one or more anycast group(s).

3.2 Protocol functionality Protocol functionality of our proposed anycast protocol can be divided into the following phases. •

•

Hello message transmission. All APs periodically transmit HELLO, which only traverse m hops (i.e., inside the proactive region), as defined by using the TTL value in the IP header. Upon receiving a Hello packet, the route to the AP is created or updated, including the current capacity of the AP, as well as the generic cost of the path to the AP. Only nodes within m − 1 hops distance from the AP decrease the TTL value and rebroadcast the packet. Route discovery (Proactive region). An MN determines that it is in the proactive region if it has received HELLO from any AP that belongs to the destination anycast group in the previous Route Expiration time interval. If the capacity of that AP can satisfy its application requirement, it can start sending data using the information in the routing table without performing route discovery; otherwise, it performs route discovery as reactive region nodes.

3.3 Analysis of optimal proactive radius Notice that the hybrid proactive/reactive approach in our hybrid anycast protocol can reduce overhead of AP discovery. However, the radius m of proactive region is an important parameter, which can greatly influence the network performance. Therefore, in this subsection, we focus on analysis of the optimal m value in terms of overhead. Before giving our theoretical analysis, we list the assumptions of our models: 1

APs and MNs have the same transmission range r

2

All APs belong to the same anycast group and their coverage does not overlap with each other

3

MNs only rebroadcast a RREQ message once

4

Both MNs and traffic sources are uniformly distributed in the network

40 5

L. Cao et al. The network is distributed in a disk area with radius R.

Therefore, the total number of overhead can be expressed as a function of m, (if HELLO and RREQ packets have the same size)

Table 1 presents all the parameters we use in our analysis.

(R2 − lm2 r2 )2 t . f (m) = l[π((m − 1)r)2 d] + πsd η R2

Table 1 Parameters in our analysis and their values in the plots

(6)

Symbols

Meaning

Values

If HELLO and RREQ are not of the same size, we can modify the above equation to:

R r m s

Network radius Transmission range of an AP or MN Radius of proactive region of an AP Number of traffic sources in the network Number of APs in the anycast group MNs distribution density Total time of operation HELLO broadcast interval

2000 m 250 m Various Various

(R2 − lm2 r2 )2 t f (m) = αl[π((m − 1)r)2 d] + βπsd , η R2

l d t η

Various 4 per m2 500 seconds 20 seconds

The total overhead of AP discovery can be divided into two parts: Hello messages from APs inside the proactive region and RREQ messages from MNs inside the reactive region. Here, we ignore the RREP messages, since they are sent along unicast routes which leads to a much lower number as compared to the number of HELLO and RREQ messages which are sent by flooding. Each HELLO message floods the proactive region and it can reach m hops with m − 1 re-broadcasts. Therefore, the total number of HELLO messages broadcast per AP is π((m − 1)r)2 d,

(1)

where d is the node density and π((m − 1)r) is the area of the proactive region of this AP. Then, the total number of HELLO messages from all l APs during the whole operation is t (2) l[π((m − 1)r)2 d] . η On the other side, we calculate the total number of RREQ messages. If the source node is located in the proactive region, there should be no RREQ overhead; otherwise, the number of RREQ messages per flooding for one route discovery is 2

π(R2 − lm2 r2 )d.

(3)

Here π(R − lm r ) is the area of the reactive area. Notice that when a RREQ reaches the proactive region of any AP in the anycast group, it will not be rebroadcast anymore. Since we assume s sources are uniformly distributed in the network, the number of traffic sources located in the reactive area is R2 − lm2 r2 s . (4) R2 Thus, the total number of RREQ messages of all the s sources can be calculated by multiplying equations (3) and (4): 2

s

2 2

R2 − lm2 r2 (R2 − lm2 r2 )2 2 2 2 × π(R − lm r )d = πsd . R2 R2 (5)

(7)

where α and β are the HELLO and RREQ packet size, respectively. Using a common setting of the parameters, as shown in Table 1, we plot the overhead function of f (m) with different combinations of l and s. Figure 2(a) shows the plot of f (m) with various numbers of sources when only a single AP is inside the anycast group. The following observations are summarised: •

total overhead increases with the number of traffic source increases, since the number of RREQ messages increases

•

all curves follow the same trend, first decreasing to the lowest point, then increasing, and merging together when r = 8, where it is completely proactive

•

the optimal value of m (where f (m) is minimised) increases as the source number increases, i.e., 3 for 10 sources, 5 for 15 sources, 6 for 20 and 25 sources; this has been confirmed later in our simulation.

Figure 2(b) shows a similar scenario with 4 APs. The trend of all curves is the same as that with single AP . Notice that the merge point of all curves shifts to 4, since m = 4 can guarantee the proactive region covers most of the network. To observe how the quantity of AP affects the overhead function, we fix s = 20 and plot the set of curves with different l values in Figure 2(c). The curves follow the similar trend (first decreasing then increasing) as shown in Figure 2(a) and (b). It is interesting that the optimal value of m decreases as the number of AP increases. In other words, as more APs belong to the same anycast group, each AP can reduce its proactive radius.

3.4 Simulation study on proactive radius We also conduct simulations with ns-2 (http://www.isi. edu/nsnam/ns) to test the performance of hybrid anycasting with different proactive radii. Our simulations use the IEEE 802.11 Distributed Coordination Function (DCF) MAC protocol. 100 nodes are randomly distributed in a 2200 m × 1000 m rectangular region. While 4 APs are fixed around the four corners, MNs

41

Multiple-metric hybrid anycast protocol Figure 2 Overhead vs. proactive radius: plot of f (m): (a) 1 access point; (b) 4 access points and (c) 20 sources (see online version for colours)

seconds), varying from high mobility (low pause time) to very low mobility (high pause time and almost static). Constant Bit Rate (CBR) sources are used, and the communication pairs are randomly chosen over the network. 10, 20 and 30 sources are used to represent different load degrees, each sending 4 packets per second with size 512 bytes. Figure 3 shows the simulation result, comparing the performance of our hybrid protocol with different proactive radii (0, 1, 2, 3 and 4). Due to space limitation, here we only present the delivery ratio and overhead measurements for four types of mobility. It is clear that average delivery ratio increases as pause time increases, also as the number of traffic sources decreases. In Figure 3(d), the delivery ratio of 10 traffic sources are lower than that of 20. This is because the network is nearly static when the pause time is 400 seconds, and there might be a partition which causes lower delivery ratio in the network. Although there are fluctuations in the plots, the main trend is coherent, increasing to a peak point, then decreasing, which means that the proper proactive radius selection can improve the performance, i.e., in Figure 3(b), the optimal proactive radius is 2. Figures in the lower row of Figure 3 compare the normalised routing overhead of the same scenarios, which also increases as the pause time increases. This set of curves also show a rough trend, decreasing to the lowest point, then increasing, similar to the observation in the theoretical analysis in Section 3.3. For example, in Figure 3(d), with 10 traffic sources, overhead achieves the lowest point when the proactive radius is 1, 20 sources with 3, and 30 sources with 4. This also confirms one of our conclusions from the theoretical analysis: optimal proactive radius increases as the number of traffic sources increase. It is not possible to determine a particular radius value, as it varies for different situations. One of our future work is to devise algorithms for optimal radius determination at run time based on network parameters.

3.5 Application requirements Classical unicast routing protocols aim to discover the minimum cost path to the single destination specified by the source node at network layer, for instance, AODV uses hop counts as the route metric. However, for heterogeneous access networks, anycast routing paradigm are required to locate the best AP among a group as well as discover the minimum cost path to that AP which can fulfill the various application requirements. Here, we classify the application requirements into two categories: • move freely with a maximum speed of 20 m/s using Random Way Point (RWP) mobility model (Johnson and Maltz, 1996). Each round of simulation runs for 500 seconds. We generate various mobility degrees with different pause time values (0, 100, 200, 300, 400

Requirements for AP selection. Different applications might have different requirements for certain types of resource at APs. On the other side, due to device or technology diversity, different APs might have different levels of capability. Taking capacity for example, APs utilising 3G access

42

L. Cao et al.

Figure 3 Average delivery ratio (the left column) and normalised overhead (the right column) with different proactive radii, number of sources, and pause time values: (a) pause time − 0 sec; (b) pause time − 200 secs; (c) pause time − 300 secs; (d) pause time − 400 secs; (e) pause time − 0 sec; (f) pause time − 200 secs; (g) pause time − 300 secs and (h) pause time − 400 secs (see online version for colours)

technology can support simultaneous transmissions to several users by assigning different channels to them. Therefore, the system capacity, can be modelled as maximum number of connections. On the other side, AP with 802.11 access technology

only provides one channel to all users by applying Distributed Coordination Function (DCF) MAC protocol, which uses a contention algorithm to provide access to all traffic. Thus, the current traffic load can be used as an indication for the capacity

43

Multiple-metric hybrid anycast protocol of the AP, the higher the traffic load, the less available resource, vice versa. To guarantee the quality of communication, AP selection must satisfy certain application requirements, i.e., selected AP has enough resource for the application. Notice that there are various ways to define or measure the resource at APs. For simplicity, we only use the number of connections at APs as an example of the metric for AP selection. However, other metrics (such as the AP selection metrics defined in Vasudevan et al. (2005), Chen et al. (2006) and Abusubaih et al. (2006)) can also be used in our multi-metric anycast protocols. •

Requirements for path selection. As the multihop relay service is provided by other MNs in the network, which are usually resource constrained, for example, constrained battery capacity, limited buffer space, CPU processing capability, memory size, etc. As availability of such resources can greatly affect the performance of the connection, applications may require the route discovery to guarantee the sufficiency of certain resource on the selected route. Besides the availability of resources, applications may also require to minimise certain cost metric to optimise the performance, e.g., hop count. On the other side, the ‘relay’ service is not free as it causes resource consumption at mobile users providing the service. Therefore, in order to communicate efficiently, applications might also limit certain cost metrics related to the expenditure, e.g., energy consumption. Thus, applications require using of multiple metrics for path cost calculation to guarantee the performance. Here, various route metrics at different layers (such as physical layer or link layer) can be used. There might be multiple APs that can satisfy the requirements for AP selection, and even with only one AP, there might be multiple paths available between source and the AP(s). Based on the multiple-metric path cost specified by the application requirement, path with the minimum cost value will be selected as the best route.

3.6 Multiple metrics path cost Since applications require simultaneously using multiple metrics to determine the path cost, we use a simple linear combination of different routing metrics, as shown in following equation:  αi × metrici (8) cost = cost + ∀i

where cost is the accumulated cost of previous nodes along the path; metrici is scaled value from (0, 1); and αi is the weighted factors (or called coefficients) for metrici to calculate the cost. Based on application requirement, these weighted factors can be flexibly varied to change the importance of the cost metrics during route discovery.

In Alkahtani et al. (2006), the authors proposed to apply Analytic Hierarchy Process (AHP) (Saaty, 1980) for the calculation of combined four QoS metrics. Even though AHP can normalise the value of metrics from (0, 1) based on relevant costs among paths, it requires route discovery message to carry the multiple metrics. This causes more control overhead as well as space cost at nodes to maintain the path information. Moreover, the complexity of the calculation is higher. Thus, in our design we uses the linear combination. However, with simple modification (additional fields in RREQ and AHP implementation at APs and its proactive regions), AHP can also be applied to our anycasting method. Furthermore, in heterogeneous networks, different mobile devices might have different levels of constraint for the resources, for example, laptop might have more powerful process capability compared to a cell phone. In order to fit the device diversity for heterogeneous networks, the protocol can also adaptively change the weight value based on the node class during the route discovery phase. Device classes could, for example, be differentiated based on the battery type, the amount of memory, or the air interfaces present in a particular type of mobile device.

4 Protocol implementation In this section, we briefly introduce some detailed designs in the implementation of our proposed protocol.

4.1 Packet format Four types of control packets are designed for the protocol, as explained in this section. •

Hello packet(HELLO). This packet is a special type of packet generated only by the APs, which is broadcast periodically inside the proactive region. As shown in Figure 4, the packet includes following fields: •

Type: 1-byte field keeps the packet type.

•

Life time: 1-byte field keeps the time (in seconds) during which this information is valid.

•

Originator Anycast Group ID: 2-byte field which represents the AP’s anycast group.

•

Originator IP Address: 4-byte unicast address of the AP.

•

Originator Sequence Number: 4-byte sequence number of the AP.

•

Path Cost Metric: 4-byte value which can include a set of path cost values along the path between the AP and the last hop node.

•

Current Capacity: 4-byte value keeps the current available capacity of the AP.

44 •

•

•

L. Cao et al. Route Request Packet (RREQ). For MNs that do not have any valid route available to any member of the anycast group in its routing table, RREQ packet is generated to initialise the route discovery. Figure 5 shows the format of the packet, which is similar to that of AODV protocol. The major differences are: instead of using unicast address as destination address, the packet has the anycast group ID as the destination address; two more fields are added for adapting application requirements and utilising multiple metrics as path cost. Accumulative path cost is the generic path cost to current node using equation (1). Generic application requirements includes both requirements for AP selection (e.g., capacity) and requirements for path selection (e.g., the weighted factors of each routing metrics). Route Reply Packet (RREP). This packet is generated by APs or MNs in proactive region for corresponding RREQ packets. The format of the packet, as shown in Figure 6, adds two more fields compared to that of AODV. While destination anycast group ID represents the anycast group that the destination node belongs to, the accumulative path cost is the accumulative cost along the path from the destination node to the source node. Route ERROR Packet (RERR). The route error packet (RERR) is the same as that of AODV protocol.

Figure 4 Hello packet

Figure 5 Route request packet

4.2 Cost metrics We assume that different APs may have different capacity, i.e., they can only serve up to certain amount of traffic. We use the total possible data rate at each AP as the measurement of its capacity. In our simulation, for

Figure 6 Route reply packet

a simple demonstration, our protocol adopt three path cost metrics: hop count, energy cost, and traffic load. Traffic load: We use the interval time between receiving two data packets to estimate the traffic load of a node. To do that, each node maintains a weighted interval value intval which is scaled and limited in a range of [0, 1], where 1 indicates no traffic load and 0 indicates high load. Upon receiving a data packet, intvl is updated by intvl = (1 − β) × intvlold + β × intvlnew . Here intvlold /intvlnew are the old/new interval values and β is an adjustable parameter (in our simulation, β = 0.2). Then the traffic load is 1 − intvl. Energy cost: We consider the transmission power at each node as the power cost metric. A simple power consumption model is used where the power consumed by a one-hop link uv is proportional to c||uv||4 . Here ||uv|| is the distance between nodes u and v, and c is a constant. Notice that remaining power capacity or more complex power consumption model can also be used as a power cost metric, but here we only use the simple transmission power to demonstrate the efficiency of our combination metrics for energy aspect. In NS2, the power consumed by the maximum transmission range 250 m is 0.66 watt per second, so the transmission power consumed by a link uv is 0.66 × ||uv||4 /(250)4 . Therefore, the range of power consumed each hop is from 0 to 0.66, and we can easily normalise this metric by dividing it by 0.66. Therefore, the path cost equation becomes: cost = cost + α1 × 1 + α2 × load + α3 × energy_cost. (9) Here, ‘1’ is the hop count; load represents the traffic load at the current node; and energy_cost denotes the normalised energy cost for the link from the previous hop to the current node. Different applications can define their requirement by including different sets of weighted values in RREQ. For example, an application might only want to consider energy consumption, thus, (α1 , α2 , α3 ) = (0, 0, 1). In our protocol, the path cost metric field in HELLO contains the three cost metrics; while the accumulative path cost field in RREQ contains the value of combined path cost from equation (9). In RREQ, the generic application requirement field includes two portions: one is the required capacity (data rate), and the other has three weighted factors for path cost calculation.

Multiple-metric hybrid anycast protocol

5 Performance evaluation In this section, we conduct several sets of simulations with NS-2 (http://www.isi.edu/nsnam/ns) to evaluate our proposed multiple-metric hybrid protocol. Our simulations use the IEEE 802.11 Distributed Coordination Function (DCF) MAC protocol. In order to demonstrate how different requirements and path cost metrics guide route discovery and resource consumption, we conduct simulations with three different network deployment. Following metrics are used to evaluate the performance: •

Packet delivery ratio: ratio of the data packets received at destinations to that sent out from sources

•

Average end-to-end delay: average time between data packets sent out from sources and received at destinations

•

Normalised routing overhead: ratio of control packets transmitted to data packets received at destination

•

Energy: total energy consumption at the end of the simulation

•

Throughput: the number of data packets received at each destination per second.

45 In the first set of simulations, the capacity of APs are both 12 packets per second, thus, each AP can only handle one connection. We conducted two simulations with and without requirement for AP selection, while hop count is used as the cost metric for path selection in both simulations. The simulation results show that: if the application specified capacity requirement, i.e., only APs with enough capacity can correspond RREP, the first connection chooses AP0 (since Path 1 is shorter than Path 2), then the second connection chooses AP1 since AP0 does not have enough capacity; the packet delivery ratio is 100%. Otherwise, if the protocol does not include capacity requirement as selection criteria, both connections choose AP0 ; the packet delivery ratio is only 77.67%. The throughput of the two APs are shown in Figure 8. As we can see, the low delivery ratio is caused by traffic overflow at AP1 . Figure 8 Throughput of our protocol with (top) and without (bottom) requirements for AP capacity in the first set of simulations in Scenario 1 (see online version for colours)

5.1 Scenario 1: AP selection and different cost metrics We first use a simple network with two APs to demonstrate the selection of APs based on AP capacity. The network deployment is shown in Figure 7, where black triangles are APs, blue dots represent relaying MNs, red dots represent source MNs who generate the traffics, and circles around APs are their proactive regions. Two APs AP0 and AP1 belong to a single anycast group. There are two CBR traffic sources in the centre of the network, both of which can connect to the two APs through multihop routes. It is easy to see that Path 1 is the shorter path with only 4 hops and Path 2 has less energy consumption. The data rate is 8 packets per second for both sources. The two connections start at 10 and 15 second, respectively. We set proactive region radius as 2, therefore, both of the source nodes are located in reactive region. Here, we conduct two sets of simulations to demonstrate the route discovery with different application requirements. Figure 7 Network deployment for Scenario 1 (see online version for colours)

After demonstrating how requirements for AP selection affect the route discovery, let us study how different path cost metrics affect the simulation results. In the second set of simulation, the capacity are increased to 20 packets per second for both APs. Therefore, both of the APs have enough capacity

46

L. Cao et al.

to process two connections simultaneously. Here, we ran three simulations with different requirements for path selection. As explained in Subsection 4.2, different requirements are specified by different sets

Figure 9 Throughput of our protocol with different path cost metrics in the second set of simulations in Scenario 1: (a) Hop count; (b) Traffic load and (c) TX power (see online version for colours)

of weighted values. In this set of simulation, we use (α1 , α2 , α3 ) = (1, 0, 0), (0, 1, 0), (0, 0, 1), respectively. Thus, the protocols use hop count, transmission power, and traffic load as application requirement, respectively. We compared the delivery ratio (DR), average end to end delay (Delay), normalised overhead (Overhead), and total consumed energy (Energy), as shown in Table 2. The throughput of the two APs are shown in Figure 9. As we can see, the difference of delivery ratio and overhead among the three simulations are not significant. In the simulation which used hop count as requirement, both of the connections choose path 1 (as shown in Figure 9(a)), which is shorter than Path 0. Thus, this simulation achieved the lowest delay compared to other two simulations, 10–20% lower. On the other side, since Path 2 has a smaller energy consumption, in simulation using transmission power as requirement, both sources choose Path 2 to forward traffic (as shown in Figure 9(c)). Thus, this simulation has the minimum energy consumption, 11–18% less. Also, as a result of the longer path, it has the highest delay. The performance of the simulation using load as requirements is between the above two, since the first connection chooses Path 1, and the second connection chooses Path 2 as shown in Figure 9(b). Table 2 Performance comparison of protocols utilising different path cost metrics in Scenario 1

DR Delay Overhead Energy

Hop count

Load

Tx power

100% 0.028 0.211 7.329

100% 0.030 0.203 6.708

99.83% 0.035 0.208 5.978

5.2 Scenario 2: Combining multiple metrics In the second scenario, we consider combining multiple metrics to find a path to an AP. As shown in Figure 10, the network includes one AP, five CBR traffic sources, as well as multiple paths from the sources to the AP. The sources start generating traffic one by one, and the beginning point is (10, 15, 20, 25, 30) seconds, respectively. The data rate of sources is 4 packets per second, and the capacity of the AP is 25 packets per second which is sufficient for serving all traffics. Figure 10 Network deployment of Scenario 2 (see online version for colours)

47

Multiple-metric hybrid anycast protocol We conducted four simulations with following sets of weighted values: (1, 0, 0), (0, 1, 0), (0, 0, 1) and (1/3, 1/3, 1/3) which represent hop count, load, transmission power, and combined metric of three metrics. In order to study the route selection with these different requirements, we recorded the number of packets traversed through every intermediate node in reactive region. The result is shown in Figure 11. Clearly, when using hop count, the protocol mainly chooses MN 10 to get connected to the proactive region, which makes the route shortest. When using transmission power, the packets are mainly forwarded through MNs 3 and 4, which can guarantee the minimum energy consumption. When using accumulative load as path selection criteria, the protocol tends to distribute traffic among multiple paths. However, as a result of the accumulative manner, it favours path with smaller hop count if two paths carry the same amount of load. Therefore, the traffic through MN 10 is still high. On the other side, if multiple metrics are used as path selection criteria, the protocol can more evenly distribute the traffic to some extent. Based on above analysis of path selection, it is easy to understand the result of the performance metrics shown in Table 3.

The distance between sources and AP0 is smaller than that between sources and AP1 . We vary APs’ capacity and the data rate of traffic sources to conduct three sets of simulations with this scenario. The configuration of these parameters are listed in Table 4. Within each simulation set, there are eight simulations that use various requirements to guide the route discovery, as shown in Table 5. For example, in Simulation 1, the application informs routing agent its data rate (as shown in the first row of Table 4) as well as the weighted values, which is (1, 0, 0); the second simulation used the same weighted values, however, it does not include the requirement for AP selection. The simulation results of Set 1 are shown in Table 6, and the throughput of each AP is shown in Figure 13. Obviously, the capacity of both APs is enough to hold Figure 12 Simple network deployment in Scenario 3 (see online version for colours)

5.3 Scenario 3: More general case Now we consider a more general complex network as shown in Figure 12, which includes two APs located on the left and right side. There are 5 CBR traffic sources that start generating traffic one by one, and the beginning point is 10, 15, 20, 25, 30 seconds, respectively. Figure 11 Traffic distribution among paths in reactive region in Scenario 2 (see online version for colours)

Table 4 AP capacity and data rate of traffic source in Scenario 3

Set 1 Set 2 Set 3

AP0

AP1

Src1

Src2

Src3

Src4

Src5

25 15 10

25 15 50

4 4 4

4 4 5

4 4 8

4 4 4

4 4 5

Table 5 Requirements for AP and path selection in Scenario 3

Simulation Simulation Simulation Simulation Simulation Simulation Simulation Simulation

1 2 3 4 5 6 7 8

Requirements for AP selection

Requirements for path selection

Data rate NULL Data rate NULL Data rate NULL Data rate NULL

(1, 0, 0) (1, 0, 0) (0, 1, 0) (0, 1, 0) (0, 0, 1) (0, 0, 1) (1/3, 1/3, 1/3) (1/3, 1/3, 1/3)

Table 6 Performance comparison of simulations in Set 1 of Scenario 3

Table 3 Performance comparison of protocols utilising different path cost metrics in Scenario 2

DR Delay Overhead Energy

Hop count

Load

Tx power

M-Metric

99.68% 0.030 0.298 3.864

99.69% 0.032 0.301 3.716

99.38% 0.038 0.37 2.673

99.67% 0.034 0.328 3.245

Simulation Simulation Simulation Simulation Simulation Simulation Simulation Simulation

1 2 3 4 5 6 7 8

DR (%)

Delay

Overhead

Energy

99.84 99.84 99.52 99.52 99.52 99.52 99.84 99.84

0.034 0.034 0.035 0.035 0.040 0.040 0.038 0.038

0.399 0.399 0.372 0.372 0.442 0.442 0.393 0.393

7.05 7.05 6.555 6.555 5.617 5.617 5.832 5.832

48

L. Cao et al.

all the traffic load, therefore, the requirement for AP selection does not take effect, i.e., the results are the same for the protocols that use the same weighted values, e.g., Simulation 1 and Simulation 2. As we can see from Figure 13(a), nearly all the traffic in the first simulation

are directed to AP0 (except several single packets at the beginning of the connections), which has the smaller cost in hop count to the sources. We can observe the similar trend in Figure 13(e), because the energy cost is less to connect to AP0 as compared to that of AP1 . This

Figure 13 Throughput of simulations in Set 1 of Scenario 3: (a) Simulation 1;(b) Simulation 2; (c) Simulation 3; (d) Simulation 4; (e) Simulation 5; (f) Simulation 6; (g) Simulation 7 and (h) Simulation 8 (see online version for colours)

49

Multiple-metric hybrid anycast protocol can also be verified from the performance metrics in Table 6: Simulation 1 achieves the smallest delay while Simulation 5 obtains the minimum energy consumption. In Simulation 3, as shown in Figure 13(c), AP0 got connected with three traffic sources while AP1 got two. As we have explained in Scenario 2, accumulative load metric prefers shorter paths. The throughput of the APs in Simulation 7 is similar, three sources chose AP0 while the other three chose AP1 . The simulation result of the second set is shown in Table 7, and the throughput of the APs is shown in Figure 14. In this set of simulation, we have the same data rate on the source side, but we decrease the capacity to 15 for both APs. Therefore, at most, each AP can only accommodate 3 connections simultaneously. In Simulation 1 and 5 (Figure 14(a) and (e)), we observe that connections are distributed across two APs. Once AP0 reaches its maximum possible capacity, new connections are directed to AP1 . In Simulation 2 and 6 (Figure 14(b) and (f)) , as there is no capacity requirements on APs, all the connections are formed with AP0 which has smaller cost. However, this reduces the overall delivery ratio as shown in Table 7. Simulations 3, 4, 7 and 8 (Figures 14(c), (d), (g) and (h)) depict similar behaviour as in Set 1. Interestingly, the figures for load metric or combined metric with and without requirement for AP selection are almost the same. This shows just using load metric or combined metric can spread the traffic among different APs. In the third set of simulations, we consider heterogenous AP capacities and various traffic demands. The simulation result is shown in Table 8, and the throughput of the APs is shown in Figure 15. Here, the data rate for the five traffic sources are 4, 5, 8, 4 and 5 packets, in the order of the start time of traffic. The capacity is configured as 10 and 30 packets for AP0 and AP1 , respectively. Therefore, unlike previous simulation settings (Sets 1 and 2), the connection distribution is different due to the different data rates and AP capacities. In Simulations 1 and 5 (Figure 15(a) and (e)), AP0 only accepts the first two connections. Simulation 2 (Figure 15(b)) has the similar trend as compared to that in Set 2, except that the upper bound is decreased to 10 packets. In Simulation 6 (Figure 15(f)), due to the heavy traffic load along the path to AP0 which causes MAC layer collisions, the last connection is directed Table 7 Performance comparison of simulations in Set 2 of Scenario 3

Simulation Simulation Simulation Simulation Simulation Simulation Simulation Simulation

1 2 3 4 5 6 7 8

DR (%)

Delay

Overhead

Energy

99.84 83.66 99.52 99.52 100 83.06 99.84 99.84

0.036 0.034 0.035 0.035 0.040 0.041 0.038 0.038

0.377 0.477 0.372 0.372 0.422 0.53 0.393 0.393

6.927 7.05 6.555 6.555 5.767 5.617 5.832 5.82

towards AP1 . In Simulation 3 (Figure 15(c)), with the capacity requirement, the connections are directed as follow: sources 1 and 4 are connected to AP0 while sources 2, 3 and 5 are connected to AP1 . Simulation 7 (Figure 15(g)) follows the same trend. Without limitation of capacity requirement, it is easy to understand the trend in Figure 15(d) and (h) for Simulation 4 and 8. Table 8 Performance comparison of simulations in Set 3 of Scenario 3

Simulation Simulation Simulation Simulation Simulation Simulation Simulation Simulation

1 2 3 4 5 6 7 8

DR (%)

Delay

Overhead

Energy

99.88 47.49 99.49 91.29 100 56.86 99.75 78.88

0.037 0.036 0.046 0.04 0.045 0.046 0.040 0.039

0.284 0.622 0.301 0.342 0.316 0.579 0.303 0.395

8.836 9.306 8.697 8.42 7.744 7.124 8.309 8.327

In summary, our simulations show different requirements and routing metric can affect the AP selection and routing performances. Our multiple metric anycast routing provide flexibility of picking appropriate metrics for applications and users. On the other hand, AP capacity should be considered in the AP selection phase to guarantee the performance.

6 Conclusion We presented an anycast protocol for heterogeneous access networks to enable MNs to select one of multiple eligible access points. We further integrated a hybrid proactive and reactive approach to AP discovery, which significantly reduces the communication overhead. The theoretical analysis shows that the selection of the proactive radius can affect the network performance and thus an optimal proactive radius is derived. We also conducted a set of simulations to evaluate the performance of the protocol, and as an extension, we utilise traffic load as the cost metric for AP selection, and let APs dynamically adjust their attitudes for dissemination of HELLOs and acceptance of RREQs regarding their load states. The simulation results show that our protocol effectively improves the performance and provides the provision for load balancing and high service availability. To satisfy various application requirements (both AP capacity and path metrics), we used combined multiple-metric to guide the route discovery and AP selection. A set of simulations has been conducted to evaluate the performance of the proposed protocol. Simulation results showed that the utilisation of multiple-metric is important and effective to guide route discovery and AP selection in heterogeneous wireless access networks. Further research work is in progress for building of a dynamic algorithm to increase and decrease AP flooding

50

L. Cao et al.

Figure 14 Throughput of simulations in Set 2 of Scenario 3: (a) Simulation 1; (b) Simulation 2; (c) Simulation 3; (d) Simulation 4; (e) Simulation 5; (f) Simulation 6; (g) Simulation 7 and (h) Simulation 8 (see online version for colours)

Multiple-metric hybrid anycast protocol Figure 15 Throughput of simulations in Set 3 of Scenario 3: (a) Simulation 1; (b) Simulation 2; (c) Simulation 3; (d) Simulation 4; (e) Simulation 5; (f) Simulation 6; (g) Simulation 7 and (h) Simulation 8 (see online version for colours)

51

52

L. Cao et al.

radius, based on network conditions. Also, further analysis can be done on the effect of heterogeneous transmission ranges of devices, as well as coverage overlapping of APs.

Acknowledgment The work of Yu Wang is supported in part by the US National Science Foundation (NSF) under Grant No. CNS-0721666, CNS-0915331, and CNS-1050398, and by Tsinghua National Laboratory for Information Science and Technology (TNList).

References Alkahtani, A.M.S., Woodward, M.E. and Al-Begain, K. (2006) ‘Prioritized best effort routingwith four quality of service metrics applying theconcept of the analytic hierarchy process’, Computers & Operations Research, Vol. 33, 3 March, pp.559–580. Abusubaih, M., Gross, J., Wiethoelter, S. and Wolisz, A. (2006) ‘On access point selection in IEEE 802.11 wirelesslocal area networks’, Proc. 6th International Workshop on Wireless Local Networks, November, Tampa, FL, pp.879–886. Aggelou, G. and Tafazolli, R. (2001) ‘On the relayingcapability of next- generation GSM cellular networks’, IEEE Personal Communications, Vol. 8, February, pp.40–47. Akl, R. and Park, S. (2005) ‘Optimal access point selection and traffic allocation in IEEE 802.11 networks’, Proc. 9th World Multiconference on Systemics, Cybernetics and Informatics (WMSCI 2005): Communication and Network Systems, Technologies and Applications, Vol. 8, July, pp.75–79. Bhargava, B., Wu, X., Lu, Y. and Wang, W. (2004) ‘Integrating heterogeneous wireless technologies: a cellular aided mobile ad hoc networks’, ACM Mobile Network and Applications, Vol. 9, No. 4, August, pp.393–408. Cao, L., Sharif, K., Wang, Y. and Dahlberg, T. (2009) ‘Multiple-metric hybrid routing protocol for heterogeneous wireless access networks’, Proc. 6th IEEEConsumer Communications and Networking Conference, Las Vegas, NV, pp.1–5. Chen, J., Chen, T., Zhang, T. and Berg, E. (2006) ‘WLC19-4: effective AP selection and load balancing in IEEE802.11 wireless LANs’, IEEE Global Telecommunications Conference, San Francisco, CA, pp.1–6. Hassanein, H.S. and Zhou, A. (2001) ‘Routing with loadbalancing in wireless ad hoc networks’, Proc. 4th ACM International Workshop on Modeling, Analysis and Simulation of Wireless and Mobile Systems, Rome, Italy, pp.89–96. Hou, Y., Shi, Y. and Sherali, H. (2005) ‘On base station selection for anycast flow routing in energy-constrained wireless sensor networks’, Second International Conference on Quality of Service in Heterogeneous Wired/Wireless Networks, Florida, August.

Hsieh, H-Y. and Sivakumar, R. (2002) ‘On using the ad-hoc network model in cellular packet data networks’, MobiHoc ‘02: Proceedings of the 3rd ACM International Symposium on Mobile Ad Hoc Networking & Computing, June, Lausanne, Switzerland, pp.36–47. IEEE Standard (2008) ‘IEEE Standard: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications Amendment 1: Radio Resource Measurement of Wireless LANs’, IEEE Std 802.11k2008 (Amendment to IEEE Std 802.11-2007), 12 June, pp.c1–222. Intanagonwiwat, C. and Lucia, D. (1999) The Sink-Based Anycast Routing Protocol for Ad Hoc Wireless Sensor Networks, Department of Computer Science, USC, Tech. Rep. Johnson, D. and Maltz, D. (1996) ‘Dynamic source routing in ad hoc wireless networks’, Mobile Computing, Vol. 353, pp.153–181. Kim, D., Garcia-Luna-Aceves, J., Obraczka, K., Cano, J. and Manzoni, P. (2002) ‘Power-aware routing based on the energy drain rate for mobile ad hoc networks’, Proc. IEEE International Conference on Computer Communication and Networks, October, Miami, FL, pp.565–569. Kubisch, M., Mengesha, S., Hollos, D., Karl, H. and Wolisz, A. (2003) ‘Applying ad hoc relaying to improve capacity,energy efficiency, and immission in infrastructurebased WLANs’, Proc. Communications in Distributed Systems (KiVS), Germany, February, pp.195–206. Lakkavalli, S., Negi, A. and Singh, S. (2003) ‘Stretchable architectures for next generation cellular networks’, International Symposium on Advanced Radio Technologies(ISART), March. Lee, J., Han, B., Lim, H., Kim, Y., Saxena, N. and Chung, T. (2009) ‘Optimizing access point allocation using genetic algorithmic approach for smart home environments’, The Computer Journal, Vol. 52, No. 8, Oxford University Press, pp.938–949. Lee, S.J. and Gela, M. (2001) ‘Dynamic load-aware routing in ad hoc networks’, Proc. of IEEE Int’l Conf. on Communications (ICC), Vol. 10, June, Helsinki, Finland, pp.3206–3210. Lenders, V., May, M. and Plattner, B. (2006) ‘Densitybased vs. proximity-based anycast routing for mobile networks’, 25th International Conference on Computer Communications, Barcelona, Spain, April, pp.1–13. Liu, B., Liu, Z. and Towsley, D. (2003) ‘On the capacity of hybrid wireless networks’, Proceedings of INFOCOM: 22nd Annual Joint Conference of the IEEE Computer & Communication Societies, Vol. 2, San Francisco, CA, April, pp.1543–1552. Peng, G., Yang, Y. and Gao, C. (2005) ‘ARDSR: an anycast routing protocol for mobile ad hoc network’, Proceedings of the IEEE 6th Circuits and Systems Symposium on Emerging Technologies: Frontiers of Mobile and Wireless Communication, Shanghai, China, Vol. 2, June, pp.505–508. Perkins, C., Belding-Royer, E. and Das, S. (2003) Ad Hoc On-Demand Distance Vector Routing, IETF RFC 3561. Perkins, C.E. and Royer, E.M. (1999) ‘Ad-hoc on demand distance vector routing’, Proceedings of IEEE WMCSA’99, New Orleans, LA.

53

Multiple-metric hybrid anycast protocol Perkins, C. and Bhagwat, P. (1994) ‘Highly dynamic destination-sequenced distance-vector routing (DSDV) for mobile computers’, ACM Conf. on Communications Architectures, Protocols & Applications, October, London, UK, pp.234–244. Saigal, V., Nayak, A.K., Pradhan, S.K. and Mall, R. (2004) ‘Load balanced routing in mobile ad hoc networks’, Computer Communications, Vol. 27, 3 February, pp.295–305. Saaty, T.L. (1980) The Analytic Hierarchy Process: Planning, Priority, Setting and Resource Allocation, 1st ed., McGraw-Hill, Columbus, OH. Singh, S. and Raghavendra, C.S. (1998) ‘PAMAS-power aware multi-access protocol with signaling for ad hoc networks’, ACM SIGCOMM Computer Communication Review, Vol. 28, 3 July, pp.5–26. Singh, S., Woo, M. and Raghavendra, C.S. (1998) ‘Poweraware routing in mobile ad hoc networks’, Proc. MobiCom, October, Dallas, TX, pp.181–190. Thepvilojanapong, N., Tobe, Y. and Sezaki, K. (2005) ‘HAR: Hierarchy-based anycast routing protocol for wireless sensor networks’, The 2005 Symposium on Applications and the Internet, Italy, February. Toh, C.K. (2001) ‘Maximum battery life routing to support ubiquitous mobile computing in wireless ad hoc networks’, IEEE Communications Magazine, Vol. 39, No. 6, June, pp.138–147. Vasudevan, S., Papagiannaki, K., Diot, C., Kurose, J. and Towsley, D. (2005) ‘Facilitating access point selection in IEEE 802.11 wireless networks’, IMC ‘05: Proceedings of the 5th ACM SIGCOMM Conference on Internet Measurement, October, Berkeley, CA, pp.293–298. Wang, J., Zheng, Y. and Jia, W. (2003) ‘An AODVbased anycast protocol for mobile ad hoc network’, 14th IEEE Proceedings on Personal, Indoor and Mobile Radio Communications, Beijing, China, Vol. 1, September, pp.221–225.

Wu, K. and Harms, J. (2001) ‘Load-sensitive routing for mobile ad hoc networks’, Proc. IEEE International Conferenceon Computer Communication and Networks, October, Scottsdale, AZ, pp.540–546. Wu, X., Chan, S-H. and Mukherjee, B. (2000) ‘MADF: a novel approach to add an ad-hoc overlay on a fixed cellular infrastructure’, IEEE Wireless Communications and Networking Conference, Chicago, IL, Vol. 2, September, pp.549–554. Wu, H., Qiao, C., De, S. and Tonguz, O. (2001) ‘Integrated cellular and ad hoc relaying systems: ICAR’, IEEE Journal on Selected Areas in Communications, Vol. 19, No. 10, October, pp.2105–2115. Yanikomeroglu, H. (2002) ‘Fixed and mobile relaying technologies for cellular networks’, 2nd Workshop on Applications and Services in Wireless Networks, France, July, pp.75–81. Ye, Z., Krishnamurthy, S. and Tripathi, S. (2003) ‘A framework for reliable routing in mobile ad hoc networks’, Proceeding of INFOCOM: 22nd Annual Joint Conference of the IEEE Computer & Communication Societies, Vol. 1, April, pp.270–280. Zhou, J. and Yang, Y. (2002) ‘Parcels: pervasive ad-hoc relaying for cellular systems’, Proceedings of MedHocNet, Sardegna, Italy, September.

Website The network simulator – ns-2, http://www.isi.edu/nsnam/ns/