Phase 1: 31 Jun 2012 - 31 Dec 2012
During the initial exploration phase, we setup the tools and investigated conceptual building blocks for MPTCP mobility
- optimization proposals for mobility aware MPTCP
- analysis of various congestion control policies for MPTCP: coupled, olia
- investigation of kernel awareness to interface appearance/disappearance
- investigation of Android, Linux differences
- mobility analysis for L2, L4
Design efforts, and development results obtained on Linux and Android platforms:
- installation of MPTCP patched kernel on Linux mobiles, server
- installation of MPTCP patched kernels on Android (Samsung Nexus) devices
- placing MPTCP proxy at CS department, in UPB
- 3G support for MPTCP in Android Intel devices
- experimental evaluation of MPTCP driven handoff between 3G and WiFi in Bucharest subway
This is a promising result as it shows that with typical urban intermittent connectivity obtained
from a combination of ubiquituous 3G coverage with patchy WiFi coverage, MPTCP can provide both continuous
connectivity, and substantial higher throughput in the wild.
- Internal dissemination inside Intel: upstream propagation of MPTCP Android related code
- Internal Intel conference on Android Development
- Presentation at Intel Academic Open Day
- submitted papers to Mobicom 2013, COnext 2013
Phase 2: 1 Jan 2013 - 31 Dec 2013
During phase two we began our work into reducing energy consumption of mobile devices and ensuring a seamless mobile experience. The three main work items in this period were:
- Design and implementation of the Mobile Kibbutz, a system where nearby mobiles pool their cellular links to save energy and reduce their network latency. See below for details.
- Initial implementation of channel switching support in the Linux kernel. This is the first part towards ensuring a true mobile experience at layer 2.
- Implementing 3G support for the Samsung XE700T Android tablet (Intel chipset), that will be used for some of the mobility tests in the future.
The mobile kibbutz
Mobile devices today rely on cellular (3G or LTE) connectivity because it has ubiquitous coverage. Unfortunately cellular links are energy hungry at low bit-rates and have high round-trip times after idle periods. These characteristics punish common mobile applications such as web browsing and streaming, decreasing battery life and user satisfaction.
We present the design and implementation of the Mobile kibbutz, a system that improves the performance of cellular links by having nearby users share their links with each other via shorter range wireless links - WiFi or Bluetooth. Without requiring application specific knowledge, instrumentation, or recompilation, the kibbutz allows multiple users' traffic to be consolidated on a subset of links, leading to reduced energy consumption and smaller round-trip times, while ensuring fairness across different users.
The concept is presented in the figure above. The black user is listening to Internet radio, while the red user is browsing the web. In isolation, both users keep their cellular links busy most of the time (Fig. 3.a), despite the low data rate. In addition, the red user experiences high delays when accessing every single page because his link must transition from idle to the connected state. By consolidating traffic onto a single link we can leave the other link IDLE all the time; this reduces the total energy consumption to approximately half. Additionally, the red user's delay decreases because state transitions are unnecessary most of the time.
We have implemented the kibbutz on Samsung Galaxy Nexus devices and shown that for typical usage applications, collaborating phones consume 25% less energy when listening to radio, experience more responsive web browsing (1.2s faster on average), and 33% faster downloads.
- Romanian patent application for the Mobile Kibbutz.
- Our poster titled "Mobile Kibbutz for Faster, Cheaper Cellular Connectivity" has appeared at Mobicom 2013.
Phase 3: 1 Jan 2014 - 31 Dec 2014
During phase three we continued our work into reducing energy consumption of mobile devices and providing seamless mobility. The three main work items in this period were:
- Design and implementation of Wifi mobility support that uses the emerging Multipath TCP standard. Here we focused on single channel mobility (in addition to multi-channel mobility reported in the previous phase)
- Energy consumption optimizations for video streaming (youtube).
- Work towards upstreaming the MPTCP kernel.
Towards Wifi Mobility without Fast Handover
WiFi is emerging as the preferred connectivity solution for mobile clients because of its low power consumption and high capacity. Dense urban WiFi deployments cover entire central areas, but the one thing missing is a seamless mobile experience. Mobility in WiFi has traditionally pursued fast handover, but we argue that this is the wrong goal to begin with. Instead, we propose that mobile clients should connect to all the access points they see, and split traffic over them with the newly deployed MPTCP protocol. We let a mobile connect to multiple APs on the same channel, or on different channels, and show via detailed experiments and simulation that this solution greatly enhances capacity and reliability of TCP connections straight away for certain flavours of Wifi (a/b/g). We also find there are situations where connecting to multiple APs severely decreases throughput, and propose a series of client-side changes that make this solution robust across a wide range of scenarios.
The solution is conceptually very simple, and is shown in the figure above: we have the client associate to multiple APs, obtaining one IP address from each, and then rely on MPTCP to spread data across all the APs, with one subflow per AP. As the mobile moves, new subflows are added for new APs, while old ones expire as the mobile loses association to remote APs.
- Our Kibbutz paper has appeared in the CONEXT 2014 conference. You can download it in PDF form here
- Our poster titled "Towards Wifi Mobility without Fast Handover" has appeared at NSDI 2014.
Phase 4: 1 Jan 2015 - 31 Dec 2015
MPTCP in enterprise WiFi networks
We have implemented a suite of modifications to ns-3 to allow simulation
of mobility inside a network with many access points under the same coordination.
These modifications include: standard 802.11 active scanning (scanning for AP-s on all
channels, recording their signal strength), triggering the active search based on different conditions
missed beacons, or beacons with low signal strength. A simulation showing throughput under
handoff conditions restricting the power of received beacons is shown below.
Energy consumption optimization on Android
An MPTCP proxy to optimize energy consumption on a device using 3G and WiFi alternatively based
on the instantaneous energy cost of the transfer. Another situation in which savings can be achieved
is when the screen has high consumption relative to the wireless/3G interfaces,
and the best startegy is to use both interfaces to race to the deadline. The architecture of the system is shown below.
Multipath Client Traffic Balancer
A Multipath Client Traffic Balancer (CTB) is a module that resides inside the ISP allowing the operator
to control the way in which clients balance their traffic across their devices. CTB achieves this by
manipulating the subflows of the MPTCP flows. The most common usecase is when the client uses
4G and WiFi interfaces towatrds a proxy that is inside the ISP. The figure below shows the general architecture of the CTB.
If you have questions about the Mobil4 project, please contact Costin Raiciu,
the project coordinator, at