Energy Consumption

Finally, NOPPoS radically reduces power consumption when compared to WLAN- Opp. Considering neighbor discovery, NOPPoS effectively replaces Wi-Fi scanning, utilized by WLAN-Opp in IDLE and STA state, with Bluetooth LE inquiring and inquiry scanning (beacon transmission and reception). The following values are cal- culated based on the specifications of an LG Nexus 5X. It incorporates the QCA6234 integrated dual-band Wi-Fi and Bluetooth 4.0 combined chip [3]. The relevant power

4 Neighborhood-Aware Opportunistic Networking on Smartphones

0 2 4 6

long medium short

average rest time

Number of AP tr

ansitions per h

NOPPoS WLANOPP

Figure 4.12: The number of AP transitions for various rest times of the random trip mobility model 0 20 40 60 80

long medium short

average rest time

Time spent in ST

A state in %

NOPPoS WLANOPP

Figure 4.13: The time spent in STA state for various rest times of random trip mobility model

4 Neighborhood-Aware Opportunistic Networking on Smartphones 0 5 10 15 20

long medium short

average rest time

Time spent in AP state in %

NOPPoS WLANOPP

Figure 4.14: The time spent in AP state for various rest times of the random trip mobility model

Table 4.4: Wireless energy consumption of QCA6234

System Continuous RX (mW) Continuous TX (mW)

Wi-Fi 2.4 GHz 54 Mbps 227.7 824.9

Wi-Fi 5 GHz 54 Mbps 247.5 989.9

Bluetooth Inquiry 0.873 39.6

consumption specifications are depicted in Table 4.4.

The continuous combined power consumption of Bluetooth results in 40.5 mW. A typical Wi-Fi scan on Android mobile devices takes 4,5 s of the 5 s scan interval leading to a RX duty cycling of 90 % or 222.75 mW at 5 GHz. This means NOPPoS reduces power consumption while in discovery phase by 87 %. This is achieved by only activating Wi-Fi scanning once the Bluetooth LE discovery phase yields possible neighbors in AP mode. Avoiding to open empty access points more efficiently also helps reducing the power consumption in AP mode by over 20 % depending on the node density (Figure 4.10).

4 Neighborhood-Aware Opportunistic Networking on Smartphones

4.6 Conclusion

This section introduced NOPPoS, a neighborhood-aware opportunistic networking ap- proach on smartphones. As major contribution, NOPPoS is governed by refined equa- tions for the state transitions at each node. These equations are based on knowing the exact number of other nodes in the radio range of each node. It was shown that the number of other nodes in the neighborhood can be accurately determined by periodic, low-energy scans. Therefore, NOPPoS is both highly responsive to node mobility and energy-efficient.

NOPPoS was evaluated utilizing the Haggle mobility trace. The presented quan- titative results evidently show that NOPPoS outperforms the approach WLAN-Opp. NOPPoS creates larger groups than WLAN-Opp, nodes spend less time in AP state. Thus, energy consumption is further reduced and the contact utilization ratio is in- creased by up to 33%.

5 Applications for Opportunistic

Gigabit Networks

This chapter builds on ideas of the NOPPoS protocol from Chapter 4. First, parts of NOPPoS are refactored and tightly coupled with a cross-layer optimized document sharing application to create Neighborhood Document Sharing (NDS). Afterwards, a recommendation system is described that can be utilized to retrieve interesting documents automatically using NDS.

5.1 Neighborhood Document Sharing

The first proposed new application for IEEE 802.11 based P2P networks is a proximity- based document transfer protocol. It is called Neighborhood Document Sharing, short NDS. In contrast to the opportunistic networking scheme in the previous chapter, a cross-layer solution is developed that tightly couples the network and the application. It enables users to discover and retrieve arbitrary documents shared by other users in their proximity, i.e. in the communication range of their IEEE 802.11 interface. However, IEEE 802.11 connections are only used on-demand during file transfers and indexing of documents in the proximity of the user. This saves energy and minimizes the use of the IEEE 802.11 interface for only high-throughput operations.

In contrast to widely available solutions like Airdrop, documents are not pushed from sender to receiver. The user has a coarse overview of all documents available in his proximity, independent of the users that are sharing the documents. Essentially, only the user that is retrieving documents is interacting with the application and downloads interesting documents.

Just like in NOPPoS, Bluetooth LE is employed additionally to the Wi-Fi interface to discover other NDS devices. It is used to broadcast the device status (e.g. NDS

5 Applications for Opportunistic Gigabit Networks

role and device id) to neighbors and to communicate the WPA2 pre-shared key and SSID to join the network. This is done by creating specially crafted Bluetooth LE advertisements with proprietary GATT services.

5.1.1 Related Work

Additional to the related work of NOPPoS, NDS touches another field of research. NDS can be seen as an implementation of the Information-Centric Networking (ICN) paradigm where devices are only interested in retrieving selected content. All requests on network layer only focus on content. NDS describes the components to implement an ICN on off-the-shelf smartphones.

Hail, Amadeo, Molinaro and Fischer [34] proposed caching and content forwarding strategies that could be easily implemented in NDS to extend NDS to a complete ICN system.

Lindemann and Waldhorst [49] proposed in 2004 Passive Distributed Indexing (PDI). PDI caches the broadcasted queries of all interconnected devices in an IEEE 802.11 network. However, NDS aims to minimize the IEEE 802.11 connections between mo- bile devices and utilizes special device roles to manage the index. Broadcasts are not required.