5G Era: Say Goodbye to TCP/IP

I just don't love you anymore, a song that can be played on repeat for a week, now sounds like 5G singing to TCP/IP.

As we all know, 4G LTE is fully IP-based, and the upper layer transmission uses the TCP/IP protocol.

However, this familiar architecture does not realize the full potential of mobile networks and even hinders the development of future networks.

LTE User Plane Protocol Architecture

TCP/IP restricts mobile Internet

About a year ago, I heard an expert complain that TCP/IP was a mistake from the beginning...

The Internet and TCP/IP are complex political, economic, social and technological products, with a lot of motivations behind them that cannot be explained by technology alone. They are using the banner of "openness" to crush the "closed" telecommunications empire, which is unfair and wrong.

Why does TCP/IP have the upper hand? The Internet world itself is a product of trial and error, and TCP/IP is no exception. However, TCP/IP dominates the Internet era in a totalitarian manner, which may lead us to make more mistakes in the future.

There is no doubt that the TCP/IP protocol has connected countless computers and promoted the booming development of the Internet since its invention in the 1970s.

However, the Internet was originally designed for fixed networks and network interconnection, but today we have entered the era of mobile Internet. Looking forward, 5G will face various applications such as AR/VR, ultra-high-definition video, Internet of Things, and Internet of Vehicles. There are many use cases, and the urgency of network security is becoming more and more prominent. The TCP/IP protocol group is difficult to adapt to the future, and we need to redefine new protocols.

The mistakes we made over the years

Indeed, 3GPP and IETF are both great standardization organizations.

3GPP's cellular mobile network has changed the world unprecedentedly. Global roaming and ubiquitous mobile connections have brought a wonderful mobile life to mankind. GSM is the greatest mobile voice communication network to date, and GPRS, UMTS and LTE have pushed us into the mobile Internet era. Mobile networks have become an indispensable part of people's daily lives, just like water and electricity.

IETF also provides network protocols for mobile networks, including TCP/IP, SCTP/DIAMETER (for LTE S1 access and core network signaling), and SIP (for VoLTE).

It can be said that the combination and complementarity of the two have promoted the prosperous era of mobile Internet.

However, this combination is a bit "forced". The Internet is designed for fixed networks and network interconnection, while 3GPP was born for mobile connections; Internet connections are distributed, while mobile communication networks are centrally controlled. The two are fundamentally different, which inevitably leads to problems such as network inefficiency and high costs.

To explain this issue in detail, we have to start with the protocol architecture of 3GPP LTE.

For the user side,

LTE continues GPRS and UMTS and adopts the IP-based communication protocol - GTP tunnel protocol. It establishes a direct GTP tunnel from end to end, and the user IP is transmitted from the terminal (UE) to the core network (PGW) through the GTP bearer tunnel. Under this architecture, the access network is ignored, there is no user-level routing selection, no ability to perceive content, and no inherent user plane security mechanism. Even every mobility migration has to update the GTP tunnel.

This end-to-end user plane architecture requires many protocols to "bridge" layer by layer, including GTP tunnel encapsulation and decapsulation, IPSec encryption, ROHC compression, etc. This undoubtedly increases transmission latency and processing costs.

For the control surface,

GTP-C is a signaling control protocol. The establishment of EPC bearer requires a separate GTP-C control protocol. Only when the EPC control plane bearer establishment signaling is completed can IP packets be transmitted between the mobile network and the external Internet.

This bearer establishment method from the control plane to the user plane also increases transmission delay and processing cost.

So, we keep making mistakes.

From GPRS, UMTS, to LTE, we have spent a lot of effort and huge amounts of money to invest in wireless access in order to improve wireless speeds. However, we have not made any progress in the protocol architecture. As part of the end-to-end ecosystem of mobile Internet, we need a more efficient mobile network to enable every part of the network to play its due role.

This inefficient GTP&IP combination wastes our wireless investment and urgently requires us to redefine new protocols.

There is also the issue of security.

Safety***

The current mobile Internet has no network access security mechanism based on the users themselves.

SSL, TLS, and HTTPS are all one-size-fits-all, end-to-end encryption methods that add overhead to all packets and span a wide range of applications and services.

However, users in the 5G era will be more picky and applications will become more diverse. Online payment, Internet of Things, Internet of Vehicles... Different application scenarios have different security requirements, some require APP-level security, and some require network-level security.

Users have higher security requirements, and all security must be considered at the user level, rather than a unified security mechanism. As an operator, we have to keep a close eye on user security issues in real time. The problem is that if there is no user-level security management mechanism for network user plane access, then we will inevitably spend a lot of manpower and material resources to ensure user security, which will increase network OPEX.

Therefore, we need a more flexible and future-oriented security mechanism. For example, we may choose user-plane-based access authentication instead of control-plane-based user authentication.

In short, looking to the future, the current protocol architecture of mobile networks has restricted development. 5G is our biggest opportunity. Otherwise, we may have to wait another 10 years.

If we do not improve the protocol architecture from base stations to the Internet, the traditional inefficient GTP&IP will inevitably waste more of our wireless investment, and the network's performance, security, mobility and scalability will be restricted, which will further affect operators' expansion of new businesses such as AR/VR and the Internet of Things, and the expansion of vertical market.

It's time for a change.

Therefore, major organizations have proposed the Next Generation Protocol Group (NGP) initiative, including the Telecommunications Standards Institute (ETSI) and the Internet Engineering Task Force (IETF) to replace TCP/IP.

Perhaps, the future 5G protocol architecture will be like this...

The song started again...

I just don't love you anymore

I don't love anymore

No more love

Never love again