Long-Term Evolution (LTE) is the realization of a longstanding aspiration of the wireless industry: a flat, all-IP (Internet protocol) access technology that’s optimal for the delivery of many types of user traffic. Voice service, however, is an exception due to its latency-sensitive nature.
The challenge is that IP traffic delivery is performed on a “best-effort” basis. While generic Voice over IP (VoIP) is a feasible low-cost solution in many instances, its inability to offer guaranteed service quality has led the wireless industry to focus on delivering “carrier-grade” or “telco-grade” voice services on the all-IP LTE network, namely Voice over LTE (VoLTE).
The network backbone that makes VoLTE possible is the IP Multimedia Subsystem (IMS) and its associated Session Initiation Protocol (SIP). This combination ensures not only that VoLTE works with other similar devices, but that the VoLTE-capable device also can communicate with non-VoLTE devices such as landlines and legacy wireless phones. Yet VoLTE quality of service (QoS) requires more than just IMS and SIP. The implementation of features in the LTE radio access network (RAN) also plays a critical role in differentiating between VoLTE and “best-effort” VoIP.
The burden on the world’s cellular infrastructure continues to grow rapidly. While advanced modulation schemes and antenna techniques have made great strides in increasing spectral efficiency, usable radio spectrum is finite and expensive. If voice traffic were treated as “generic” packet data, an LTE network would deliver it (alongside every other type of traffic) via physical downlink and physical uplink shared channels (PDSCH/PUSCH) using the default evolved packet system (EPS) bearer originally designed to deliver generic data traffic to an application server such as a Web server.
However, “best effort” Internet/IP delivery affords no control over service quality. LTE introduces the concept of an EPS dedicated bearer to establish specific tolerances for packet delay, error loss, and guaranteed versus non-guaranteed bit rates. For controlled QoS for VoLTE traffic, the system sets up an EPS dedicated bearer dedicating to transporting voice packets (Fig. 1).
Multiple users share the physical-layer (PHY) channels (PDSCH and PUSCH). Data is multiplexed by dividing the channels into resource blocks (RBs). Each RB is a grouping of 12 subcarriers in one time-slot duration. An RB is the minimum allocation of the LTE PHY resource that can be granted to the user equipment (UE).
Since every RB on the downlink and uplink must be granted, VoLTE introduces a challenge: the overhead in granting RBs becomes too great for the persistent and near continuous RB allocation required for the delivery of the small packets that are typical of a VoIP-based conversation. Semi-persistent scheduling (SPS) minimizes this overhead by taking advantage of the predictable transmission pattern associated with small VoLTE packets. SPS establishes a consistent pattern of RB grants rather than requiring a separate grant for each 20 ms of audio. A radio resource control (RRC) message controls the periodicity of the recurring RB grant (Fig. 2).
Robust Header Compression
VoLTE involves several protocol layers. Each has associated overhead, creating the potentially inefficient scenario in which overhead messaging occupies more data bandwidth than the encoded VoLTE data itself. In the case of IPv6 the combination of RTP, UDP and IP headers can be 40 to 60 bytes long.
Robust header compression (RoHC) takes advantages of redundant header information to greatly reduce overhead. As an example, the 40 to 60 bytes of header associated with an IPv6 VoLTE packet can be reduced to as little as 3 to 4 bytes.
New features of the LTE RAN take advantage of the relatively low data bandwidth requirements of voice audio. SPS takes advantage by pre-scheduling the periodic allocation of resource blocks. RoHC minimizes the bandwidth that would otherwise be used to transmit redundant header information at multiple protocol layers.
Discontinuous reception (DRX) takes further advantage at the radio layer. VoLTE encodes 20-ms chunks of audio conversation and then rapidly burst-transmits the encoded audio. A period of no transmission follows each encoded audio packet transmission. During these silent periods, DRX turns off the UE’s RF receiver, analog-to-digital converters, and digital signal processors to reduce battery drain and increase talk and standby usage time.
Transmission Time Interval Bundling
In wireless communications, the transmission time interval (TTI) is the time needed to transmit the smallest unit of data allowed by the technology. This is 1 ms in LTE, compared to 2 ms in HSPA and 10 ms in WCDMA. Since resource scheduling is done on a per-TTI basis, this smaller TTI facilitates low latency, which is an advantage when delivering latency-sensitive data such as voice.
However, this shorter TTI can cause issues at cell edges. Without TTI bundling, the hybrid automatic repeat request (HARQ) process requires an 8-ms interlace period before it can detect an error and request that a packet is re-encoded and re-transmitted. TTI bundling pre-supposes that there will be errors on the transmission (Fig. 3). Rather than waiting for the HARQ mechanism to trigger a re-transmission, TTI bundling sends redundant copies of the same data, encoded with different error detection/correction bits, within a single HARQ period.
LTE Voice And Legacy Voice Services
With the industry-wide recognition that large-scale rollouts of this technology can’t be rushed, it will take some time for LTE coverage to become as ubiquitous as the underlying 3G coverage. In the interim, subscribers must be able to transition voice calls to a legacy network when roaming out of LTE coverage.
One way to seamlessly transition a VoLTE call to a legacy network is called Single Radio Voice Call Continuity (SRVCC). SVRCC not only promotes call continuity, it’s also more cost effective than current solutions, including circuit switched fallback (CSFB) and simultaneous voice and LTE (SVLTE), which requires multiple radios within a UE.
However, considering that the UE must tune to a new frequency/band, acquire and transmit on the legacy network, and transition from the packet-switched to circuit-switched delivery, SRVCC implementation is itself a complex phase in VoLTE evolution. This complexity is expected to delay commercial release of SRVCC until 2013 at the earliest, which is likely a factor in the decision by some larger operators to delay their commercial implementation of VoLTE.
Each of these RAN and mobility features requires support within the UE. A robust UE test plan requires an emulated network that provides complete integration of IMS and evolved packet core (EPC) infrastructure, multiple radio access technologies, and the ability to efficiently control IMS messaging so the emulated network behaves identically to a live cellular network.
Additionally, RRC procedures must be efficiently configurable to establish and terminate SPS, enable and disable DRX, turn TTI bundling on and off, and implement RoHC. For example, all these design verification test (DVT) capabilities are available today in Spirent’s CS8 Device Tester. This benchtop solution offers an interactive mode of operation for debugging issues or creating ad-hoc test scenarios as well as visual and programmatic interfaces for creating automated tests applicable to every phase of the UE design cycle.
The successful implementation of VoLTE is a critical part of the cellular industry’s future, especially network operators and mobile device manufacturers. Besides the well-known IMS required to make VoLTE a reality, the LTE RAN introduces features specifically designed to make VoLTE an efficient reality.These RAN features have a direct effect on achieving successful VoLTE QoS. Ensuring these features work as intended is critical, and due to the underlying complexities, testing must be achieved in a timely and cost-effective manner.