The IP’s of Mobility
Last month, the article Mobile Wireless Systems and Evolution focused on the evolution of mobile wireless networks from circuit-switched, voice-centric technologies and architectures to IP Next Generation (IP NGN), packet-switched, data-centric architectures where voice was nothing more than an overlay network.
Continuing in the series of articles related to mobility, this month’s article (part 2 of 4) will focus on how mobility is instrumented in mobile networks today, and discuss future considerations within IETF for handling mobility in the future.
Mobility is a term commonly used, but seldom do two people, or even worse, two standards, use the same definition or understanding of mobility. Mobility is commonly grouped into sub-categories, which leads to further confusion. For instance:
Seamless mobility vs. Nomadic mobility: Seamless mobility refers to the ability to change a network attachment point while the session is in-process without losing connectivity. Nomadic mobility refers to the ability to change network attachment point while the session is stopped.
Micromobility vs. Macromobility: “Micro” and “Macro” are often used to refer to the scale of mobility. Micromobility tends to deal with situations where a subscriber is seamlessly moving between two points of attachment that are part of the same network. Macromobility tends to deal with situations where a subscriber is seamlessly moving between networks (regardless of operational control).
In addition, mobility can also be categorized as follows:
Terminal mobility: Terminal mobility refers to either micro- or macro-mobility cases for a specific device, or terminal. The majority of mobile service providers today offer some level of terminal mobility.
Personal mobility: Personal mobility refers to the subscriber’s, or person’s, ability to move between network attachment points AND between devices while still accessing services. Unified Communications provides some level of personal mobility with the capability for value-add services such as single number reachability, where a single phone number can ring multiple devices.
Session mobility: Session mobility refers to the ability to hand off in-session services between different devices. Unified Communications may also provide session mobility through call transfer, where a voice call can be handed off from a landline to a mobile device without dropping the call.
The “Layers” of Mobility
The most logical way to represent mobility, however, for those of us who have a long technical history in the IP world, is through a layered approach mapped to the OSI model. The figure below illustrates the mapping of many of today’s commonly-known protocols into the OSI model.
The sections that follow discuss each of these layers, and, in particular, the protocols that map to these layers. The figure below provides a framework for how the layered mobility model relates to the categorization of terminal, personal, and session mobility.
Link Layer Approaches to Mobility
Link layer approaches to mobility are designed to handle terminal mobility without affecting higher layers. In other words, changes to the lower layer are invisible to the upper layer, and vice-versa. These approaches and protocols are have historically been used to manage micro-mobility cases (i.e., point of attachment might change, but network-layer information remains intact).
General Packet Radio Service (GPRS) Tunneling Protocol (GTP) is perhaps one of the most widely deployed mobility technologies in the world. Developed as part of 3GPP for use in GPRS, UMTS, HSPA and LTE networks, GTP provides mobility services for over 80% of the world’s population.
Lightweight Access Point Protocol (LWAPP) is designed to provide lightweight mobility between WiFi access points in an enterprise or WiFi mesh deployment. LWAPP operates at either Layer 2 or Layer 3, and preserves general routability to/from a WiFi client by encapsulating their traffic in either an Ethernet frame or UDP packet.
Proxy Mobile IP (PMIP), or network-based mobility management, provides a mechanism for mobility that does not require any modification to the subscriber device network stack. Like other link layer approaches, PMIP allows the subscriber to move points of attachment without requiring a new IP address, and therefore, neither the subscriber nor the corresponding device is aware that any mobility happened at all. PMIP is defined as a protocol for both WiMAX and LTE networks for mobility.
Network Layer Approaches to Mobility
Network layer approaches to mobility are designed to handle terminal mobility in a loosely-coupled model between higher layers and lower layers. These protocols rely on network-layer functions for mobility.
Mobile IP (MIP) Version 4 (MIPv4) and Version 6 (MIPv6) rely on encapsulation to tunnel traffic between two points in the network, known as a Foreign Agent (FA) and Home Agent (HA). The Foreign Agent is responsible for recognizing when a subscriber has moved network boundaries and notifying the Home Agent of the new local IP address of the subscriber. The Home Agent is responsible for presenting a globally-static route to the corresponding nodes for a particular subscriber. There are many “flavors” of MIP that have been developed to handle inter-access and inter-technology mobility, including Dual Stack MIP (DS-MIP), Simple Mobile IP (SMIP), Fast Handovers for Mobile IPv6 (FMIP), and many others designed to enhance and adapt Mobile IP protocol to changing requirements.
Host Identifier Protocol (HIP), Location Identity Split Protocol (LISP), and 8+8 all fall under the category of “location-identity splitting protocols.” These protocols leverage the large address space allocated for IPv6 nodes, and split the address into two portions – the upper bits as the “locator,” and the lower bits as the “identifier.” The locator is a dynamic portion of the address that changes as the subscriber moves between network attachment points. The identifier is a fixed portion of the address that uniquely identifies the subscriber device. When put together, the full address notifies a correspondent of subscriber@location in order to adequately route traffic to the subscriber. These location-identity splitting protocols are promising areas of research that may potentially provide tunnel-less approaches in the mobile Internet.
Upper Layer Approaches to Mobility
Upper layer approaches to mobility are defined at the application, session, and transport layers. These protocols manage mobility either through application-layer signaling or socket-level functions. In general, these protocols are designed to provide session continuity across devices and technologies, although a number of working groups within IETF are attempting to provide this same functionality at the network layer. These groups, such as Mobility Extensions for IPv6 (MEXT), Network Based Mobility Extensions (NETEXT), are driving the next-generation of network mobility standards.
Stream Control Transfer Protocol (SCTP) provides a similar role in the network as that of Transmission Control Protocol (TCP), including reliable transport, sequencing, and congestion control. In addition, SCTP provides a multi-homing capability, which allows both endpoints in a communication to use multiple IP addresses. From a mobility perspective, this feature of the SCTP protocol can allow a mobile device to stop transmitting on one IP address and begin transmitting on another IP address once session handover is complete.
MSOCKS provides a similar capability at the transport layer to what Mobile IP does at the network layer. Mobile IP relies on a proxy device to modify the IP layer so that a corresponding node is unaware that a mobile device has moved point of attachment and changed IP address. MSOCKS relies on a proxy to modify the TCP layer by splicing together two TCP connections (mobile host <-> proxy, proxy <-> corresponding host) that appears as a single, end-to-end, TCP connection.
Session Initiation Protocol (SIP) is an application-layer protocol that provides a signaling layer capable of initiating handoffs. SIP relies on a URI which uniquely identifies a subscriber on the network, and is bound to the subscriber IP address. When the mobile device changes point of attachment, and therefore IP address, the mobile node also issues a SIP Re-Invite message, which communicates either directly to the corresponding node, or to a SIP proxy, that the IP address associated with the URI has changed.
Regardless of the layer upon which mobility rides, the end-result of many completed and ongoing efforts in standards organizations will deliver a new era of communications – one in which subscribers are always connected, able to seamlessly access content and services regardless of whether they are fixed in a location, nomadic, or fully mobile. This new era will rely on many different approaches to mobility that must cooperate and coordinate in order to provide a full range of mobility options to tomorrow’s mobile subscribers.
If you have any questions regarding the topics discussed in this article, you can email questions to Kevin from now through July 30, 2009.