When the 3GPP announced the beginning of 5G development back in 2015, the group proposed many performance objectives pertaining to all aspects of the network. Eventually the 3GPP distilled the various metrics down to three distinct use cases: enhanced mobile broad band (eMBB), Ultra Reliable Low Latency Communications (URLLC), and Massive Machine Type Communications (mMTC). There has been much written and published on the eMBB use case and even the mMTC to a lesser extent with the anticipation of smarter devices and pervasive IoT. Although the URLLC use case has garnered less attention, it may in fact prove more impactful than either of the former use cases.

The pivotal term in URLLC is “low latency”. In some ways ultra reliable and low latency represent two separate and potentially orthogonal goals. As defined by Wikipedia, “..latency is defined as the time interval between the stimulation and response, or, from a more general point of view, a time delay between the cause and the effect of some physical change in the system being observed”.  

While physical layer researchers differ on the exact definition of latency,  it is generally referred to as the round trip time from when a transport block contained in a slot is sent from a base station (BTS) and the user equipment (UE) responds to the initial transmission of the transport block. This is a narrow view, but makes it controllable from a research perspective at the foundational level of the standard. All variables that impact latency in this definition can be controlled by the physical layer designers. With the pending finalization of 3GPP Release 15, the initial 5G NSA Phase 1 release, the 3GPP contributors addressed latency at the physical layer with several optimizations, which I’ll outline below.

First, the flexible numerology scheme allows for slots in a subframe (1ms) to be defined as uplink, downlink or some combination of the two. Second, the time duration of the slot is flexible and depends on the sub carrier spacing (SCS). The 3GPP specifies several SCS options dependent on spectrum and bandwidth. Each slot represents 14 OFDM symbols and each subframe can scale in time as noted by the table below:

For the URLLC case, the shorter slot durations are important. The 3GPP has also defined “mini-slots” that further reduce the timing to 2, 4, or 7 symbols and would cause the timing in the table above to scale linearly. Finally, the 3GPP also defined the “self contained” subframe case. In this mode, transmit and receive from the UE side occurs wholly within a single subframe. Self contained subframes include the HARQ  which in theory makes a significant reduction in latency possible. The HARQ timing generally increases access time depending on the quality of the link and can increase latency significantly.

Network researchers note that the physical layer and improvements in the full stack, ie the RAN, are only part of the latency equation. The data bits must be sent and received from a UE, however if the sending device is located on the other side of the world, minimal latency targets will be difficult to realize. Physics dictates the travel based on distance, and even the fastest networks must address this challenge.

Researchers refer to this type of latency as end-to-end or E2E. E2E low latency is not possible without modification to the core network and potential inclusion of network slicing and/or Mobile Edge Computing (MEC) nodes. By separating the control and user planes in the standard, the 3GPP opened the door for new network topologies to enable network slicing. More to the point, a distributed network control methodology where the network can direct a packet using the shortest path to a computational node to efficiently reduce the E2E latency is needed.

The 3GPP has laid a solid foundation for realizing lower latency in 5G networks. While the improvements in the physical layer will be realized for the eMBB and mMTC use cases, the potential for URLLC remains. However, URLLC and true low latency applications may have to wait for further definition of the upper layers in the 5G Core Network scheduled for release in December of 2018.