Mobile data offloading is a concept of shifting traffic from mobile cellular networks onto a WiFi network instead. This concept started with 2.5G networks when data began to compete with voice services and expanded with 3G and 4G networks. It is enabled by WiFi-enabled devices, such as Android and Apple smartphones, which are designed to switch automatically to WiFi when it is available, as WiFi offers better speeds and reliability than cellular networks, and at a lower cost. This requires, in many cases, the user to authenticate the connection before they can connect, and the network and the user's device can "remember" the other in the future. While in some cases, such as public WiFi, users may be required to authenticate the switch because the public network presents security risks.
However, for mobile data offloading in the strictest sense, the user does not need to do anything to access the WiFi network, as the switch and authentication happen automatically and are built into the service. Mobile virtual network operators (MVNO) may maintain their own networks and hotspots that allow for mobile data offloading without the subscriber needing to do anything. Mobile data offload has offered mobile network operators (MNOs) a chance to provide carrier-class WiFi and mobile services to customers for an improvement in service with "always-on" data connectivity, with WiFi deployments offering MNOs 30 to 40 percent savings on their network CAPEX while gaining several thousand percent in radio network capacity. Whether MNOs partner with WiFi service providers or build their own carrier-class WiFi networks, WiFi offers MNOs an opportunity to address a market for connectivity.
Mobile data offloading not only offers the mobile user the speed and connectivity of WiFi networks but also reduces cellular network congestion. Offloading is able to more cost-effectively meet the ever-growing data needs of customers and enterprises. In 2018, mobile phones accounted for 52 percent of all traffic worldwide, while the consumption of web video over cellular networks was growing at a rate of around 70 percent, year over year. Mobile data offloading is a way of reducing cellular network congestion and saving money for both the network operator and the subscribers. For these subscribers, mobile data offloading offers them the speed and reliability of the network they expect, while most experts believe that mobile data offloading is a more effective strategy for reducing network congestion than other common solutions, such as scaling and optimizing networks.
Given the limited range of WiFi access point placement, data analytics have to be used by mobile network operators to identify where to build access points for the greatest number of their subscribers for mobile data offloading. Given population density, most of these access points tend to be in dense urban areas, where the highest amount of traffic and subscribers meet. Mobile data offloading for reducing network congestion offers an opportunity for third parties to develop community networks, WiFi mesh networks, or a licensed band femtocell, which they can license to mobile network operators and their subscribers.
Mobile data offloading can help reduce the gap between the cost of the facilities needed to support mobile network traffic growth, which is seen in mobile operators investing in expanding the capacity of the Radio Access Networks (RAN) through upgrades and deploying additional cell sites by adding capacity to mobile backhaul facilities. The use of mobile data offloading and making investments into owned or leased network capacity can increase the perceived value a subscriber has in their mobile service while also improving the mobile operator's profitability.
For an effective implementation of mobile data offloading from cellular to WiFi networks, there are certain aspects that need to be addressed. The aspects tend to be fairly obvious, such as the need to increase the WiFi footprint in order to implement offloading solutions, the challenges inherent with user equipment and necessary enhancements, user authentication challenges, and inter-network mobility considerations.
The coverage offered by WiFi has to be increased to have the option for mobile data offloading, while cellular network operators require agreements with existing wireless internet service providers, if they do not have their own infrastructure, to provide traffic offloading. Which scenario the MNO chooses depends on what fits their business scenario, with the main intent of increasing the ubiquity of WiFi networks so they are readily available for offloading mobile data traffic.
Mobile data offload requires minimal or no user interaction with the network to initiate offload, with most devices handling the offload automatically and requiring network authentication at the most in many scenarios. This offers a satisfyingly seamless user experience but requires the provision of smart connection managers in the user's equipment capable of interacting with the network and engaging in flexible and efficient offloading policies. For example, the connection manager should be capable of making decisions regarding network selection, volume of traffic to be offloaded, application and service-specific offload, time-based offload management, and other challenges.
Similar to smart connection management, a user's device requires the appropriate authentication mechanisms for both cellular networks and WiFi networks, as both of these networks have different mechanisms for authentication. Techniques like SIM-based authentication can be used for WiFi networks and are becoming increasingly popular, while most implementations of how policy and charging rules are applied in offload scenarios depend on proprietary policies. Standards for the policy and charging rules have been developed and applied in offload scenarios, including the Access Network Discovery and Selection Function (ANDSF) and the Policy and Charging Rules Function (PCRF) developed by 3GPP to cater to this functionality.
As WiFi was mainly developed for local area networks and lacked mobility functionality, cellular and WiFi integrations have been recommended and have gained momentum, with the potential for seamless mobility, where users can switch from network to network without noticing. This has led the Internet Engineering Task Force (IETF) to develop specifications for the Mobile Internet Protocol (MIP) and Proxy Mobile Internet Protocol (PMIP) to address mobility issues in WiFi.
Standardization in cellular and WiFi integration began as early as 2002 when GSMA formed a group to study possible interworking scenarios for cellular and WiFi technologies, and the group's findings would result in 3GPP developing specifications for cellular and WiFi integration. These specifications are typically divided into two groups based on the mobile core network they pertain to: Universal Mobile Telecommunication System (UMTS) and Evolved Packet Core (EPC). The standards for UMTS are referred to as IWLAN (Integrated/Interworked WLAN) standards and the latter as EPC standards for non-3GPP access.
The IWLAN standards for mobile data offload in UMTS core network using WiFi cover the aspects of common billing and customer care, 3GPP system-based access control and charging, access to 3GPP system PS-based service, service continuity, seamless service and access to 3GPP CS services. One main goal of the IWLAN solution was to achieve authentication without manual user intervention common with WiFi networks and develop authentication protocols based on the use of SIM cards. SIM-based authentication is familiar for 3GPP-based network operators while providing a similar level of security as 3GPP devices. These modified authentication protocols are known as EAP-SIM (Extensible Authentication Protocol-SIM).

UMTS core architecture supporting IWLAN.
Another aspect of the IWLAN standard is mobility management to provide seamless mobility between cellular RAN to WiFi RAN and between inter-operators WiFi RANs to the user. The mobility function was developed based on an IP-level mobility management protocol called DSMIPv6, standardized by IETF. These protocols allow handover from 3GPP access to WLAN access or vice versa. The main drawback of the IWLAN standards is the offloading architecture, which is not connected to policy and quality of service management entities in the core network, which prevents advanced policy and quality of service-based management of the standard. Further, the IWLAN standard is limited to a single radio connection at a time, where it should be possible to support multiple interworking connection flows on a device at a given time.
The EPC standards for non-3GPP access aim to provide a higher level of integration between the WLAN and cellular technologies for tighter interworking. The EPC standards offered an era of trusted non-3GPP access with a view that the cellular operator would also own the WLAN network. This included enhancements on the IWLAN standards, such as the Packet Data Network Gateway (PGW) and the Policy and Charging Rule Function (PCRF) connected to various gateway functions and able to enforce operator policies. The EPC standard also makes it possible to support certain IP-Flows on the 3GPP radio interface and other levels on the WLAN radio interface, based on criteria such as quality of service requirements, user subscription, type of user equipment, and the ability to enable dynamic switching of individual IP-Flows from one radio interface to another.

EPC architecture for trusted non-3GPP access.
Mobile data offloading tends to be comprised of two main types of offloading: delayed offloading and on-the-spot offloading. In delayed mobile data offloading, the offloading occurs when the WiFi is available, but if the user moves out of the coverage area, the packets are not immediately routed through the cellular network. The transmission of data packets through the cellular network in delayed offloading will only occur if the user does not reenter the WiFi coverage area within a given timeline. Generally, an efficient mobile data offloading scheme should have a low transmission delay and high offloading efficiency.
In on-the-spot offloading, when the WiFi network is available, data transmission is done over WiFi. When it is not available, the cellular network performs the transmission of data packets. The movement between networks will not stop the transmission of data in on-the-spot offloading, offering a more seamless experience for the user, and this technique is easier to implement as smartphones are increasingly capable of using more than one interface at a time.
One proposed type of offloading is prediction-based offloading, in which three schemes have been proposed: a hybrid scheme for message replication, forecast and delay, and a utility-based scheme. The proposed schemes work to use predictive behaviors of human mobility to perform offloading of delay-tolerant data from a remote site. This would allow the network to begin a data download at a site where a user is expected to be next, based on mobility behavior analysis, and offer the user a seamless offloading experience through a handoff of data between points, with a delay during travel between those points.
The proposed schemes each use different models to predict future contact opportunities and base these models on a time series of data of connectivity over a collected period of time to better forecast future contact opportunities. The prediction-based offloading model is best suited for IoT environments, where the user in question is expected to travel between pre-determined nodes, and where a higher degree of predictability and uniformity increases the efficiency of the offload.
Also sometimes called data offloading types, there are three distinct approaches to data offloading often expressed as automatic, opportunistic, and aggressive. The automatic method is the most common method, in which the handoff is done automatically by a device connecting to the strongest signal, whether that is WiFi or cellular. Another method is "opportunistic offloading," in which a user manually connects to a WiFi network rather than using a cellular connection. And the third, "aggressive offloading," happens when a carrier routes all traffic through a WiFi network to relieve cellular network congestion.
There are three main initiations for mobile data offloading: manual, automatic or WLAN scanning initiation, and remotely managed initiation.
Manual initiation, as its name suggests, occurs when a mobile device user manually switches the network from the cellular network to a WiFi network, most often. This is also known as user-initiated offloading.
The automatic or WLAN scanning initiation occurs when the user device performs a network scan and when a known or open WiFi network is found to perform an offload procedure. WLAN scanning is also known as WiFi preferred network offload (PNO) scanning. The scans themselves are low-powered and perform at regular intervals regardless if the user device is disconnected from WiFi or if the screen is off. Scans typically happen every 60 seconds, until a saved network is found, or the screen is turned on at which point manual offload is initiated on some devices.
A remotely managed initiation approach is when a network server initiates each offloading procedure by prompting a connection manager of a specific user device. With the remotely managed initiation approach, the operator-managed is a subclass in which the operator monitors network load and user behavior, and in the case of forthcoming congestion, the operator initiates an offload procedure.
Access Network Discovery and Selection Function (ANDSF) is a key technology for enabling carriers to offload data traffic from mobile networks. ANDSF is a part of the 3GPP standard designed to assist mobile devices in discovering offload destinations, such as WiFi, WiMAX, and CDMA2000 networks. The feature was designed to provide mobile devices with information about available alternative wireless networks and enforce policies for selecting and using those networks. The standard is location- and device-specific but not intended to be responsive to real-time network conditions.
Access Traffic Steering, Switching, and Splitting (ATSSS) is a newer standard and technology for automatic network selection and intelligent convergence between mobile and WiFi services that can be developed for the mass market, despite its complexity and reliance on device support which will increase the time it can take to reach the market. The ATSSS technology is developed for 5G networks. It is developed to allow mobile operators to avoid unintentional "walk-by" switchover to public WiFi, which can reduce user experience, and set policies and thresholds that should automatically reject or accept handoff to WiFi and back to cell sites if either is congested.
The steering function allows mobile network operators to choose the best available network for a device based on speed, cost, and latency. Switching allows the device to move seamlessly between 5G and WiFi networks. And splitting allows the traffic over 5G and WiFi to be split, which can be set by policies. This technology is in part developed by the multipath capability of 5G user devices.
There are various necessary technologies to allow for mobile data offloading. These tend to be complementary network technologies that enable the handoff and allow for the mobile device user to switch between network services. WiFi services are an obvious such technology. Another is femtocells, small low-power cellular base stations that can be designed for use in a home or small business. As mobile data offloading and user usage of data continue to increase, the technologies for mobile data offloading continue to expand.
WiFi tends to be the most used mobile data offloading technology, as more devices are equipped with WiFi chipsets, allowing them to connect to both the mobile network and a nearby WiFi gateway. Mobile operators have consistently turned to WiFi as a complementary technology to work with traditional mobile access technology, leading to a rise in carrier-class WiFi networks. This is important as it ensures the end-user experience is not degraded when a mobile handset switches between access technologies. WiFi also works as many areas are already equipped with WiFi networks, such as in airports, hotels, and city centers, and WiFi tends to be less costly to build than cellular networks.
Hotspot 2.0 (HS-2.0) has been proposed to simplify the access procedure and enhance the working ability of WiFi networks, which has drawn attention from industrial manufacturers and open source communities, respectively, while also offering a chance of improving WiFi offload. This is partially done by reducing the procedure required by WiFi networks when switching from one access point to another, which HS-2.0 works to simplify through seamless and secure connectivity with automatic access that allows uses to switch from one network to another seamlessly and makes this technology more suited to offloading.
WiMAX is a technology for data offloading for various markets, implemented through WiMAX Gateways and WiFi offloading. These solutions can be integrated gateway solutions and WiFi gateways for implementing WiMAX data offloading, which can be used by industries for OEMs and SIs to integrate in order to develop network solutions. The WiMAX gateway is an integrated software solution that consists of a C ASN Gateway, an authentication, authorization, an accounting (AAA) server, and home agent components, all co-located on a single network to support nomadic and mobile subscribers. The framework is capable of scaling to support 200,000 subscribers and serve the needs of rural, Tier 3, and enterprise deployments and was developed for third-generation networks.
Femtocells are primary offload technologies used by network operators and use standard cellular radio technologies. Any mobile device is capable of participating in the data offloading process, though some modification can be required depending on the backhaul connection used. Femtocells can also be used in situations in which local cell reception is poor. Femtocells are connected to a mobile network using residential DSL, cable broadband connections, optical fibers, or wireless last-mile technologies. Femtocells are a popular option as they are easier for the mobile phone to exchange with, and they are less power intensive than WiFi offloading, offering greater battery savings for users.

Example of a femtocell-based mobile data offload architecture.
However, because femtocells use the same frequency band as macrocells, interference problems can occur, which have prevented widespread adoption of femtocells. Femtocell interference can also be between two different femtocells. This has led to the development of algorithms that can perform self-configuration to reduce power consumption and interference. This has also led to the proposal of traffic offloading between femtocells and WiFi networks using software-defined networking (SDN) technology, which allows a terminal to maintain existing sessions after offloading through a centralized control of the SDN equipment, which would work to provide seamless connectivity while also reducing femtocell load and increasing the quality of service for data throughput.
Femtocells work for a wide range of air interfaces, such as GSM, CDMA2000, TD-SCDMA, W-CDMA, LTE, and WiMAX. And because femtocells operate in a licensed spectrum, they can be managed effectively and dynamically based on best practices associated with the deployment of small cells. However, the best practices and details of these small cells change based on practical use cases and the radio technologies used. And, due to better radio conditions of cellular links between mobile users and small cells can effectively enhance the performance on data transmission.
Selected Internet Protocol (IP) Traffic Offload (SIPTO) is a 3GPP standard that allows internet traffic to flow from the femtocell to the internet while bypassing the operator's core network and relieving that network. This can relieve the EPC from high-bandwidth flows that can cause significant congestion and can identify the IP Flow type while applying a local user policy rule, which is provisioned from the mobile core network and then executes the offloading. Further, the upsides of SIPTO are that the network operator has a lower load on its network and the operator can select which traffic to offload. However, implementation of SIPTO requires a special functionality in a separate node or integrated with an existing node to be placed in the RLAN and core network to accomplish this.
Similar to SIPTO, Local IP Access (LIPA) is designed to optimize IP traffic management and focuses on IP traffic destined to a local IP network and routes such traffic instead through the mobile core network. LIPA is developed for use cases such as a home-user trying to access a local server without having access to a wide area network. LIPA is linked to femtocells implementations and standards that specify offload through home nodes. Similar to SIPTO, LIPA has limited mobility support for locally routed traffic and can use policy-driven routing to select which traffic to offload.
Integrated mobile broadcast (IMB) is a mobile wireless technology that enables the broadcast of content at the cellular transmitter level, using the licensed radio spectrum, and in turn, is capable of being received on mobile terminals. This allows customers to roam and offload seamlessly. This was developed by 3GPP Release 8 in December 2008 and has been trialed to offload bandwidth-intensive mobile data from networks and places on a broadcast portion of a licensed spectrum. Integrated mobile broadcast supports linear and non-linear broadcast services and can be implemented in TDD spectrum.
Device-to-Device (D2D) communication offers some unique advantages to offload cellular traffic, such as better system throughput, higher energy efficiency, and robustness to infrastructure failures. This was enabled in part due to initiatives from chipmaker Qualcomm to develop cross-platform communication protocols allowing for D2D communication. In D2D communication, if a popular message or data packet is being shared, and a nearby phone has a message another phone is trying to access, the neighboring device can share the message cached locally rather than retrieving it through the cellular network. However, the challenge can be the limited storage on a device and how many messages can be shared between devices. Further, in D2D communications, user devices can be used to extend a network range and "hand-over" data packets almost like in an information chain, which can also provide greater network stability, especially if the original network is already being used for mobile data offload.
Another proposed technology for mobile data offloading has been wireless mesh networks, in which a cellular-to-mesh (C2M) data offloading for cellular users can achieve offload. These would be mesh networks developed as commercial services or in context of a community network which can then be leased by mobile network operators to offload their traffic and reduce servicing costs. This approach has been proposed to be optimized for energy savings for the cellular base station based on intelligent routing policies, while also offloading traffic for the minimum possible cost.
IP flow mobility is a connection management technology in which mobile network operators can move selected IP data traffic between different infrastructures, such as between cellular infrastructure and WLAN infrastructure. All done while keeping other ongoing connections for this and the rest of users on both radio accesses untouched. This approach was adopted by 3GPP for seamless 3G offload.

Example of the mobile data offloading via IP flow mobility, in which a single process can be offloaded.
There are various ways of achieving this, such as extending client-based IP mobility solutions to allow flow mobility where the user terminal is involved in the mobility process. While a second way of achieving this is based on extending current network-based IP mobility solutions where the user terminal is not away from the mobility.
Delay tolerant networking (DTN) has been used to migrate traffic from cellular networks to free and high-capacity device-to-device networks. This works to cope with traffic demands, growth, and limited capacity demands, which can migrate cellular data traffic. DTN can be used for non-real-time applications, and service providers can delay and transfer the data for those applications until they reach less-congested networks. Providers can use DTN as a cost-effective solution for mobile data offloading while supporting the growth of mobile data.
The multipath TCP works to solve the handover problem between networks in a clean or seamless way. Multipath TCP connections are able to use both the WiFi interface and the cellular one during the handover. This allows the user device to maintain the service connectivity and maximize the usage of the cellular network while allowing for the coexistence of a cellular path and an alternative path for the same application session. Through intelligent switching between these paths, multipath TCP can determine the guaranteed bit rate and ensure the user's experience remains seamless for the intended quality of service.
Cloud computing can offer a different type of offloading, which still reduces the amount of data being passed through a network. Except, instead of offloading the data packets off of the mobile core network, the mobile computation is offloaded to the cloud, where the cloud handles the computation, meaning the majority of the data necessary for the computation is routed to the cloud rather than over the mobile network and what the phone receives is the final formulation of the computation. Further, through different algorithmic formulations and through optimizations, the computation offload occurring in the cloud can be used to balance and minimize transmission latency. For most mobile users, they will receive the data requested without realizing the computation has been passed to the cloud, except in a case where latency in the network can delay the computation results data. However, it can be argued that the same network latency would impact other offloading and data transmission, providing a near-similar experience to on-device computation.
Similar to moving device-based computation to the cloud, the core network can install small cell sites at the edges of the mobile network with the intelligence software necessary to handle the offload of data and voice traffic. This allows the core network functions to be handled at the network edge, such as switching local calls and caching popular content and allowing operators to make major savings on backhaul. This can be used for various markets such as corporations, rural areas, and remote locations while making it easier to scale up the mobile network with more edge devices rather than upgrading a central intelligence server.
To offload near the edge of the network, to mitigate the limits of network backhaul, caching near the edge of the network has been proposed. This can reduce the load on the backhaul and the core network. In one such scenario, content has been cached on user devices, where a neighboring device can request information from a device that is already storing the data, and if no such device is found the content is fetched as usual. This works differently than D2D communication, which requires the two devices to be within range at the same time and thereby require dense topologies and are only feasible with popular content. Instead, it allows one to use the mobility of nodes to increase the effective coverage of each cache, offloading considerably more data with reasonable additional delays. A second study used this model with a HetNet setup where content could be cached on small cells and devices to be retrieved locally, which could optimize the strategy.
Citizens Broadband Radio Service (CBRS) offers cost-effective LTE solutions for indoor and outdoor applications, which network operators could use as a method to offload mobile data traffic. CBRS is developed to offer network operators a high level of network quality, performance, and flexibility compared with WiFi offloading. For example, carriers with PAL licenses can use CBRS in 10-megahertz channels to allow network operators to use carrier aggregation to increase network capacity and improve data speeds. And changes by the FCC have increased CBRS deployment capabilities, allowing network operators to rely on them with more confidence.
CBRS has been suggested as being more useful, as they have been increased in their spectrum capacity, for 5G network deployments. Especially as different start-ups and companies have issued CBRS schemes, which allow individuals to establish a 5G node for a distributed network that carriers, while the user who deploys the node receives an income for traffic.
With the rollout of 5G networks, there is a vision to roll out small and constrained Internet of Things (IoT) devices to develop IoT environments. In these environments, machine-to-machine (M2M) communications are key to allow IoT devices—such as sensors, actuators, smart meters, and monitors—to collect data and direct the data to a remote server. This is achieved with a multipath TCP gateway, allowing users to connect multiple cellular networks to offload M2M data traffic, which can reduce network congestion across multiple networks for a single IoT environment, allowing for faster data transfer rates. This is partially due to the M2M data traffic offloading increasing throughput and reducing latency of CoAP requests as the number of sensor networks increases.
Other than using mobile data offloading for reducing congestion in IoT environments, IoT modules and environments can be used for mobile data offloading to reduce network congestion for mobile network operators. This would require access point owners (APOs), or the owners of the IoT modules, to be appropriately incentivized to use their networks for mobile data offloading. In this way, IoT environments could, outside of peak times, be used to reduce congestion and communication costs while continuing to earn the IoT network operator more income.
One IoT environment that continues to be developed, which could be used in this case, is vehicle-to-vehicle communication. As more vehicles are manufactured with communication modules intended developed for city-based IoT scenarios as more cities work to develop smart infrastructure. This infrastructure could be used as a device-to-device (D2D) communication strategy to reduce mobile network infrastructure, through D2D mobile data offload to the IoT network developed by vehicular networks.
There has been a belief, as 5G networks and infrastructure rollout become common, that there will not be a need for mobile data offload given the network speed and capacity. Some have even suggested 5G networks can make WiFi obsolete. However, it is more likely that 5G will be a complement to WiFi and provide a boost to speed, capacity, and quality, especially as the latest WiFi 6 technology has increased connectivity speed and capacity over traditional WiFi networks. Further, WiFi 6 is better able to handle an increased demand for data-intensive applications, while the increased capacity is also developed to help the growing IoT market, suggesting that WiFi 6 will increase the chance for WiFi-based mobile data offload. So, while 5G brings higher speeds and lower latency, it will continue to drive a need to offload traffic onto WiFi networks.
Network infrastructure for 5G is expected to require a hub-and-spoke design that moves centralized processing into areas closer to the antenna, where each spoke aligns with the antenna location, which can increase the load balancing and processing into the distributed network. This design approach is intended to offer higher-performing 5G networks and require more active networks closer to the antenna or deployed at the network's edge. And as networks are deployed and support high bandwidth and low latency, they will achieve network densification, to offer a high-quality signal for responsive mobile devices. This will allow for core-network processing and edge-computing devices to work on the network, while mobile data offloading can increase network capacity and allow 5G networks to enable different deployments based on the low-powered and low-latency network performance.