There are several papers that articulate the benefits of virtualization [7,15,4]. However, to the best of our knowledge, the benefits of virtualizing a wireless card has been overlooked. In this paper, we propose MultiNet, a new virtualization architecture that abstracts a single wireless LAN (WLAN) card to appear as multiple virtual cards to the user. MultiNet allows these virtual cards to be simultaneously connected to physically different wireless networks. We describe our architecture in detail and then present a buffering protocol and two switching algorithms that give good performance for many common applications, such as telnet, ftp, file sharing and web downloads.
Our research is motivated by several compelling scenarios that are enabled with the above functionality. These scenarios include: increased connectivity for end users; increased range of the wireless network; bridging between infrastructure and ad hoc wireless networks, and painless secure access to sensitive resources. We discuss these in detail in Section 2.
To enable these scenarios with current technology, one has to use a single WLAN card for each desired network. However, this is costly, cumbersome, and consumes energy resources that are often limited. An alternative to using more hardware is to use MultiNet and its accompanying protocols. MultiNet requires changes to the data link or device driver layer of the networking stack. It creates and manages multiple network stacks and maintains the associated state information for each network that the card is connected to. Simultaneous connectivity over all networks is achieved by switching the card between the desired networks and activating the corresponding stack. An advantage of this architecture is that it allows applications and protocols like TCP/IP to work without any changes.
In this paper we make the following four research contributions:
As of this writing, MultiNet has been operational for over twelve months. During this time, we have refined the protocols and studied their performance. Many of the results we present are based on real working systems that include current and next generation IEEE 802.11 [10] wireless cards. For cases where it is not possible to study the property of the system without large scale deployment and optimized hardware, we carry out simulation based studies. Most of our simulations are driven by traffic traces that represent `typical traffic'. For IEEE 802.11, our study shows that MultiNet nodes can save upto 50% of the energy consumed over nodes with two cards, while providing similar functionality. We also quantify the delay versus energy tradeoff for switching nodes over performance sensitive applications. Although we have built MultiNet over IEEE 802.11 wireless LANs, our approach is not limited to this standard.
The rest of this paper is organized as follows. In Section 2 we present some scenarios and applications that motivated us towards building MultiNet and for which MultiNet is currently being used. Section 3 provides the background needed for the rest of the paper and Section 4 presents some related research. The MultiNet architecture is presented in Section 5, and its implementation is described in Section 6. Performance and feasibility are discussed in Section 7 and 8. Future work is presented in Section 9 and we conclude in Section 10.