Martyn Ruks, Strategic Advisor
12 mins read
It is an exciting time in the world of telecommunications and mobile working. We have seen the unprecedented proliferation of mobile devices, including smartphones and tablets, coupled with the increasing accessibility of 4G network services. The result is a highly versatile and mobile workforce that wants to consume new technology faster than ever and wants to revolutionise their working practices and methods. So, at the point where the adoption of 4G technology by businesses and consumers alike is set to rise massively, do we understand the implications of using it? And what do providers need to do to provide assurances about its security?
First, let’s be clear about the definition of 4G. We are talking about the 3GPP Long Term Evolution (LTE) Advanced standards, and not any of the other competing technologies such as WiMAX. TheLTE standard is primarily about improving the user experience for mobile communication. However, it also includes added benefits for the operators.
In simple terms, the LTE standard aims to support the delivery of network services that:
The number of components, interfaces and protocols that existed in older methods of mobile data delivery created a barrier to deployment. The lack of geographic scalability resulted in less than optimal population coverage from network providers. The LTE specification aims to enable the delivery of improved mobile services and to reduce the complexity and cost of deploying mobile data networks.
In LTE, this has partly been achieved by consolidation of components that are used in legacy infrastructures to produce a flatter network topology. Radio layer components have been redesigned in a more modular way to allow faster deployment: this enables the effective handover of user sessions from one geographic node to another; and aspects of the non-radio layer more closely resemble traditional IP networks. These factors result in improvements to the service delivered to customers, as well as lower deployment and maintenance costs for the network provider, which should in turn act as an incentive to increase coverage.
So why should we care about security if we are going to use 4G? After all, the standards address some of the concerns raised by previous incarnations of mobile technology, so surely it must be more secure than 3G, as well as everything that came before it. I can already hear you saying “No!” After all, it uses IP for all back-end communications so it must be easier to attack and therefore must be less secure. The older technology used weird and wonderful protocols, so you couldn’t just plug your laptop into the base station and start attacking the back-end network with common tools.
When you look at LTE, you will clearly see that it uses IP networks throughout, so that you can use the tools you know and love against the back-end components. This includes the Base Station equivalent, known as the evolved NodeB (eNodeB), making it therefore much easier to attack.
If you are asked to consider the security implications of your business using a 4G network, you may feel a little bit daunted by the subject – from impenetrable acronyms and components such as the eNodeB, to the systems within the Evolved Packet Core (EPC). Additionally, components like the MME, HSS, PCRF, SGW and PGW may be new to you, particularly if you haven’t looked at a 3G network. Secondly, there are some protocols that you might not be that familiar with, for example, have you ever looked at SCTP? Do you know what S1AP is? What about GTP and all its different flavours? Putting the protocols and the components together, what would be the most profitable attacks? What could an attacker try and spoof? Where might an attacker be able to route packets to? Which protocols will they be fuzzing and what vulnerabilities might affect you as a user of the system?
Whether you are tasked with guiding your business on its use of LTE or you are a provider preparing for testing an LTE environment, there are probably a whole range of questions that you would like to be able to answer. In this knowledge centre piece, we will offer an analysis of LTE, so you can have confidence in its security. We will outline the important components in an LTE environment, scenarios for security testing that should be considered and some key security controls that should be implemented to protect the network.
The objective of this knowledge centre article is not to provide you with a full description of all the systems and protocols that comprise an LTE network. However, it does aim to provide an insight into the roles of key attributes of the system, so that you can clearly understand what is important when you come to test its security. Let’s start with some of the main components, followed by more detailed descriptions.
UE is a generic term that refers to any device or system that consumes IP services in the environment. At present, UE is primarily composed of USB dongles and LTE network hubs, but there are increasing numbers of smartphones and tablets that are 4G enabled. UE should only be capable of consuming services on the Internet or those specifically facilitated by the network operator; they should never to able to participate in direct IP communication within the environment.
An eNodeB is an evolution of the Base Transceiver Station (BTS) as present in previous GSMimplementations, and it acts as the bridge between wireless and wired networks. An eNodeB will typically have three LTE-specific interfaces, one wireless or air interface (known as Uu), one for inter-eNodeB communication (known as X2) and one for communication with the MME and Serving Gateway (known as S1). These devices may also contain other interfaces – such as those used for management – that use IP or Universal Serial Bus (USB) communication, although these are not specified by the LTE standard. The eNodeB will typically be attached to an external aerial via the Uu interface.
The EPC is the collective term for the back-end infrastructure that the eNodeBs communicate with and through which user traffic passes. The EPC contains a number of discrete components that play different roles. The primary change in these environments from what existed in previous technologies is the use of the Internet Protocol (IP) in all wired communication.
The Home Subscriber Service (HSS) is a central store of all user-related subscription data. These profiles identify the level of access that user equipment will have on the network and the services and data bearers that are mandated by these profiles. The HSS participates in the management of UE across cells, call establishment support, user authentication and authorisation. UE is authenticated to the network using data that is derived from keys that are stored within a Universal Subscriber Identity Module (USIM) and within the HSS.
The MME is the control node for the LTE network. It is responsible for the tracking and management of UE that is in idle mode. The MME is involved in the brokering of data bearers and the assignment of a Serving Gateway (SGW) to UE during the registration process. By interacting with the HSS, the MME handles authentication of UE in the registration phase.
The Serving Gateway (SGW) is primarily responsible for the management of UE state information and the routing of user data packets. Additionally, the SGW is used as an anchor point for UE crossing from one eNodeB’s coverage area to another.
The Packet Data Network Gateway (PGW) provides an entry and exit point for UE that is accessing external packet data networks. The PGW implements deep packet inspection for the profiling of data channels and the provisioning of suitable data bearers.
A unified or consolidated gateway combines the functionality of both the SGW and PGW into a single component with internal communication between the two.
One of the major changes between LTE and previous technologies is the use of the Internet Protocol (IP) for communication between components in the environment. This use of IP provides greater scope for an attacker to abuse the features of the IP protocol, specifically because the network design is more likely to share components between user and control planes.
More interestingly, a number of additional protocols use IP for their transport and require specific knowledge from any prospective tester or security monitor. The following protocols used within the wired network are critical to the security of an LTE environment and testing activities should therefore include analysis of them and the manner in which they interact with individual components:
Alongside both TCP and UDP, SCTP is used for a number of communication streams within the back-end network. The primary use of SCTP is for the handling of critical communications between eNodeBs and the MME, where robust communication is critical to the successful operation of the environment.
S1AP supports the transfer of data between eNodeBs and the MME. This protocol is used to transfer signalling information between the UE and the MME and to manage session state between eNodeBs and the MME. The protocol uses the SCTP for underlying session management and guaranteed delivery. Within S1AP, a pair of IDs is used to track the identity of an individual UE in the data that is communicated. One of these IDs is generated by the eNodeB and the other by the MME.
X2AP provides the communication of data between individual eNodeB components and is similar to S1AP in its structure. This is used to transfer information about UE when performing a mobile handover. The protocol uses the Session Control Transport Protocol (SCTP) for underlying session management and guaranteed delivery.
GTP-U is used for the transfer of user data between the eNodeB and the Serving Gateway as well as between eNodeBs during X2 handover. The protocol is used to encapsulate a user’s IP traffic so that it can be transported into the EPC where it is subsequently unencapsulated and routed onwards to its destination. As a GTP packet can be encapsulated inside another, it is possible to construct an IP packet with multiple layers of GTP data. If not correctly handled by the equipment, this feature might allow an attacker to use encapsulation to bypass security controls. The eNodeB will always add one layer of IP data to the packet sent by the UE when encapsulating the data, therefore an attacker does not have full control over its construction.
GTP-C can be used for communication between back-end components within the EPC, although it is not a principal part of the standard. The protocol is also used by legacy 3G components, although transport over IP is a requirement in LTE.
If you are running an LTE network, it is important that you have considered all of the viable attack scenarios that exist. If you are intending to use 4G, then you need to consider what the implications are if your provider hasn’t addressed them.
Attacks that are conducted across the air interface of the environment are assessed to be of the greatest concern and therefore a large amount of testing should be conducted from this perspective. However, it is important that a threat modelling based approach is used to identify where the critical controls are within any given deployment and that an appropriate level of testing is used to provide assurances about them.
There are four primary testing locations that should be considered when planning a security testing engagement within an LTE environment:
Tests conducted from this location emulate an attacker with wireless access to an LTEenvironment through an operator-provided dongle, home router or smartphone. The attacker could attack the environment through routing and spoofing attacks, primarily using IP Traffic sent from a laptop or other connected system.
The following types of testing activities are recommended at this location:
Tests conducted from this location emulate an attacker with the ability to monitor and intercept wireless communications passing between a user and an operator’s radio mast. Without access to vendor equipment, it would be very difficult to perform practical attacks at this location given the sophistication of the wireless technology that is used. However, as has been illustrated recently, it is possible (though not straightforward) to build LTE stacks that could be used to test this interface.
The following types of testing activities are recommended at this location if appropriate tools can be built:
There are controls built into the standard to provide protection against some of these techniques. However, implementation quality is still a big factor as to whether these are effective in practice.
This emulates an attacker with physical access to an eNodeB and any associated cabling or network equipment. The attacker could attempt to compromise the eNodeB physically and connect to unused ports or tamper with the network cables that are attached.
The following types of testing activities are recommended at this location:
This emulates an attacker with IP access to the network between the eNodeB and EPC and is assumed to be possible at any location between them. Conducting testing emulates an attacker who has identified a mechanism for sending and receiving traffic at this location. This could be through unauthorised physical access or through a logical attack from the air interface. At this location, the testing should include analysis of both the control plane (used for signalling) and the user plane (used for transferring a user’s data).
The following types of testing activities are recommended at this location:
There are a number of recommendations that are critical for a provider to have implemented with respect to security controls within an LTE environment. If they haven’t done so, any data traversing these networks may not be as secure as you would hope! The most important of these controls are firstly the secure design and configuration of IP routing and secondly the use of IPSec between eNodeBs and the EPC. Each of these key controls should be covered by the testing approach outlined previously, and more detail is given here.
Ensuring that UE cannot access any services within the EPC is a fundamental requirement of the security model. The design of the architecture in the core is therefore vital to achieving this effectively. Preventing the routing of traffic from UE into the inner part of the EPC should be achieved by a combination of secure design and effective routing configuration. One of the primary considerations is how traffic on the Internet-facing side of the PDN Gateway is routed, and this should ideally avoid any switches or network equipment that has a route into the core of the EPC.
The design of IP routing in the environment is complex: it will typically require the use of different types of network device; it will utilise multiple VLANs; and it will use multiple IP address ranges. If IPv4 and IPv6 support is required both for users and within the core, this can also increase complexity. A robust architecture is fundamental to securing the environment. This should be validated with security testing as described previously.
If a provider hasn’t addressed this correctly, there is the possibility that an attacker will be able to exploit these vulnerabilities and access key system components. The implications range from their being able to bypass billing through to the worst case scenario – the compromise both of the system and of the data passing through it.
In a default configuration, there is no method of providing authentication, confidentiality or integrity protection for any communication that occurs between eNodeB and the EPC. As eNodeBs will be placed in locations that may have poor physical security controls, these communications need to be secured using other means. IPSec is accepted as being the recommended method of securing communication on the S1, X2 and user plane connections within an LTE environment. However, there are several challenges in implementing it in a secure manner.
It is recommended that any IPSec connections to the eNodeB are terminated in the host, as the ability to control network level access into the EPC is vitally important. This can be achieved by terminating IPSec either at a gateway or within the individual EPC components. When you are configuring systems in the EPC and eNodeB to use IPSec, it is recommended that services and interfaces are not accessible without using IPSec. This is of particular concern when physical access can be gained to exposed interfaces on an eNodeB. IPSec also needs to be enforced on all interfaces that are enabled but not used. The quality of the IPSec implementation is another key area for security testing as described previously.
IPSec was not part of the LTE standard. However, it is required to secure the network when an eNodeB is located at an insecure location. In particular, it should be noted that the authentication that can be provided by a correctly configured IPSec implementation is not equivalent to and cannot be translated to authentication in the LTE network. A single compromised IPSec connection may allow an attacker to impersonate other nodes on the network and would expose systems to attacks at the level of the protocols that are otherwise protected within the IPSec tunnel.
If a provider hasn’t implemented this correctly, it could result in the exposure of data passing across the wired components of the network. If you were counting on the provider’s encryption to protect your data from attack, then this might be an unexpected blind-spot.
If appropriate security input is provided during the design and implementation phases of an LTEnetwork, there is no reason that a robust and secure environment should not be achieved. However, it is important that the security controls stated within the LTE standard are not blindly relied upon to protect the environment and its users. Only a good understanding of the potential risks and validation of the security controls that are implemented to mitigate them will provide assurance as to the security of an LTE environment.